2 zEnterprise Technical Details - ITSOWW2011
-
Upload
caleb-aaron-ortega -
Category
Documents
-
view
83 -
download
10
description
Transcript of 2 zEnterprise Technical Details - ITSOWW2011
yourdotcom
International Technical Support Organization and Authoring Services
© 2011 IBM Corporation
www.ibm.com/redbooks
zEnterprise Technical Details
Parwez HamidExecutive IT ConsultantIBM UK Limited
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
2
TrademarksThe following are trademarks of the International Business Machines Corporation in the United States, other countries, or both.
The following are trademarks or registered trademarks of other companies.
* All other products may be trademarks or registered trademarks of their respective companies.
Notes:
Performance is in Internal Throughput Rate (ITR) ratio based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput that any user will experience will vary depending upon considerations such as the amount of multiprogramming in the user's job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve throughput improvements equivalent to the performance ratios stated here.
IBM hardware products are manufactured from new parts, or new and serviceable used parts. Regardless, our warranty terms apply.
All customer examples cited or described in this presentation are presented as illustrations of the manner in which some customers have used IBM products and the results they may have achieved. Actual environmental costs and performance characteristics will vary depending on individual customer configurations and conditions.
This publication was produced in the United States. IBM may not offer the products, services or features discussed in this document in other countries, and the information may be subject to change without notice. Consult your local IBM business contact for information on the product or services available in your area.
All statements regarding IBM's future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.
Information about non-IBM products is obtained from the manufacturers of those products or their published announcements. IBM has not tested those products and cannot confirm the performance, compatibility, or any other claims related to non-IBM products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products.
Prices subject to change without notice. Contact your IBM representative or Business Partner for the most current pricing in your geography.
Adobe, the Adobe logo, PostScript, and the PostScript logo are either registered trademarks or trademarks of Adobe Systems Incorporated in the United States, and/or other countries.Cell Broadband Engine is a trademark of Sony Computer Entertainment, Inc. in the United States, other countries, or both and is used under license therefrom. Java and all Java-based trademarks are trademarks of Sun Microsystems, Inc. in the United States, other countries, or both. Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both.Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries.UNIX is a registered trademark of The Open Group in the United States and other countries. Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both. ITIL is a registered trademark, and a registered community trademark of the Office of Government Commerce, and is registered in the U.S. Patent and Trademark Office.IT Infrastructure Library is a registered trademark of the Central Computer and Telecommunications Agency, which is now part of the Office of Government Commerce.
For a complete list of IBM Trademarks, see www.ibm.com/legal/copytrade.shtml:
*BladeCenter®, DB2®, e business(logo)®, ESCON, eServer, FICON, IBM®, IBM (logo)®, MVS, OS/390®, POWER6®, POWER6+, POWER7, S/390®, System p, System p5, System x, System z, System z9®, System z10®, WebSphere®, z9®, z10, z/Architecture®, z/OS®, z/VM®, z/VSE, zSeries®
Not all common law marks used by IBM are listed on this page. Failure of a mark to appear does not mean that IBM does not use the mark nor does it mean that the product is not actively marketed or is not significant within its relevant market.
Those trademarks followed by ® are registered trademarks of IBM in the United States; all others are trademarks or common law marks of IBM in the United States.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
3
zEnterprise Technical Details – z114
Capacity on Demand14
Capacity and Performance Planning15
Upgrades16
Summary17
Coupling Link Connectivity and STP8
I/O Summary9
Sysplex Consisderations10
STP Support and Enhancements11
RAS13
Crypto12
Network Connectivity Features7
1 IBM System z Family
2 Model, Frame and Processor Drawer
3 Memory Design
4 I/O Infrastruture
5 I/O Subsystem Enhancements
6 Storage Connectivity Features
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
4
IBM System z Family
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
5
IBM System z10 family
�Announce 2/08 – Server w/ up to 77 PU cores�5 models – Up to 64-way�Granular Offerings for up to 12 CPs�PU (Engine) Characterization
– CP, SAP, IFL, ICF, zAAP, zIIP
�On Demand Capabilities – CoD, CIU, CBU, On/Off CoD, CPE
�Memory – up to 1.5 TB for Server and up to 1 TB per LPAR
– 16 GB Fixed HSA
�Channels– Four LCSSs– 2 Subchannel Sets– MIDAW facility– 63.75 subchannels– Up to 1024 ESCON® channels– Up to 336 FICON® channels– FICON Express2, 4 and 8– zHPF– OSA 10 GbE, GbE, 1000BASE-T– InfiniBand® Coupling Links
�Configurable Crypto Express3�Parallel Sysplex clustering�HiperSockets™ – up to 16�Up to 60 logical partitions�Enhanced Availability�Operating Systems
– z/OS, z/VM, z/VSE™, TPF, z/TPF, Linux on System z
IBM System z10 EC (2097) IBM System z10 BC (2098)
�Announced 10/08 – Server w/ 12 cores �Single model – Up to 5-way CPs�High levels of Granularity available
– 130 Capacity Indicators
�PU (Engine) Characterization– CP, SAP, IFL, ICF, zAAP, zIIP
�On Demand Capabilities – CoD, CIU, CBU, On/Off CoD. CPE
�Memory – up to 256 GB for Server– 8 GB Fixed HSA
�Channels– Two LCSSs– 2 Subchannel Sets– MIDAW facility– 63.75 subchannels– Up to 480 ESCON channels– Up to 128 FICON channels– FICON Express2, 4 and 8 – zHPF– OSA 10 GbE, GbE, 1000BASE-T– InfiniBand Coupling Links
�Configurable Crypto Express3�Parallel Sysplex clustering�HiperSockets – up to 16�Up to 30 logical partitions�Enhanced Availability�Operating Systems
– z/OS, z/OS.e, z/VM, z/VSE, TPF, z/TPF, Linux on System z
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
6
IBM zEnterprise family
�Announced 7/10 – Server w/ up to 96 PU cores�5 models – Up to 80-way�Granular Offerings for up to 15 CPs�PU (Engine) Characterization
– CP, SAP, IFL, ICF, zAAP, zIIP�On Demand Capabilities
– CoD, CIU, CBU, On/Off CoD, CPE�Memory – up to 3 TB for Server and up to 1 TB per LPAR
– 16 GB Fixed HSA�Channels
– Four LCSSs– 3 Subchannel Sets– MIDAW facility– Up to 240 ESCON channels– Up to 288 FICON channels– FICON Express8 and 8S– zHPF– OSA 10 GbE, GbE, 1000BASE-T– InfiniBand Coupling Links
�Configurable Crypto Express3�Parallel Sysplex clustering�HiperSockets – up to 32�Up to 60 logical partitions�Enhanced Availability�Unified Resource Manager�Operating Systems
– z/OS, z/VM, z/VSE, z/TPF, Linux on System z
�Announced 7/10�Model 002 for z196 or z114�zBX Racks with:
– BladeCenter Chassis– N + 1 components– Blades– Top of Rack Switches– 8 Gb FC Switches– Power Units– Advance Management Modules
� Up to 112 Blades– IBM Smart Analytics Optimizer Solution– POWER7 Blades– IBM System x Blades– IBM WebSphere DataPower Integration
Appliance XI50 for zEnterprise (M/T 2462-4BX)
–Operating Systems– AIX 5.3 and higher– Linux for x Blades– Microsoft Windows for x Blades
–Hypervisors– PowerVM Enterprise Edition– Integrated Hypervisor for System x
IBM zEnterprise 196 (2817) IBM zEnterprise Blade Extension (2458)
IBM zEnterprise 114 (2818)
�Announced 07/11 �2 models – M05 and M10
� Up to 5 CPs�High levels of Granularity available
– 130 Capacity Indicators�PU (Engine) Characterization
– CP, SAP, IFL, ICF, zAAP, zIIP�On Demand Capabilities
– CoD, CIU, CBU, On/Off CoD. CPE�Memory – up to 256 GB for Server
– 8 GB Fixed HSA�Channels
– Two LCSSs– 2 Subchannel Sets– MIDAW facility– Up to 240 ESCON channels– Up to 128 FICON channels– FICON Express8 and 8S – zHPF– OSA 10 GbE, GbE, 1000BASE-T– InfiniBand Coupling Links
�Configurable Crypto Express3�Parallel Sysplex clustering�HiperSockets – up to 32�Up to 30 logical partitions�Unified Resource Manager�Operating Systems
– z/OS, z/VM, z/VSE, TPF, z/TPF, Linux on System z
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
7
z114 continues the CMOS Mainframe heritage
0
500
1000
1500
2000
2500
3000
3500
1997Multiprise®
2000
1999Multiprise
3000
2002z800
2004z890
2006z9 BC
2008z10 BC
MH
z
139MHz
413 MHz
625 MHz
1.0 GHz
1.4 GHz
� z800 - Full 64-bit z/Architecture®
� z890 - Superscalar CISC pipeline� z9 BC - System level scaling
3.5 GHz
� z10 BC - Architectural extensions �Higher frequency CPU
� z114 – Additional Architectural extensions and new cache structure
� Multiprise 2000 – 1st full-custom Mid-range CMOS S/390
� Multiprise 3000 – Internal disk, IFL introduced on midrange
2011z114
3.8 GHz
4000
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
8
IBM zEnterprise 114 (z114)
Models, Frame and Processor Drawers
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
9
� Machine Type
– 2818
� 2 Models
– M05 and M10
– Single frame, air cooled
– Non-raised floor option available
– Overhead Cabling and DC Power Options
� Processor Units (PUs)
– 7 PU cores per processor drawer (One for M05 and two for M10)
– Up to 2 SAPs per system, standard
– 2 spares designated for Model M10
– Dependant on the H/W model - up to 5 or 10 PU cores available for characterization
• Central Processors (CPs), Integrated Facility for Linux (IFLs), Internal Coupling Facility (ICFs), System z Application Assist Processors (zAAPs), System z Integrated Information Processor (zIIP), optional - additional System Assist Processors (SAPs)
• 130 capacity settings
� Memory
– Up to 256 GB for System including HSA
• System minimum = 8 GB (Model M05), 16 GB (Model M10)
• 8 GB HSA separately managed
• RAIM standard
• Maximum for customer use 248 GB (Model M10)
• Increments of 8 or 32 GB
� I/O
– Support for non-PCIe Channel Cards
– Introduction of PCIe channel subsystem
– Up to 64 PCIe Channel Cards
– Up to 2 Logical Channel Subsystems (LCSSs)
� STP - optional (No ETR)
z114 Overview
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
10
� M/T 2818 – Model M05
– Air cooled– Single Frame– Non-raised floor option available– 30 LPARs
� Processor Units (PUs)– New processor drawer design (1
processor drawer)– 7 per system
• 2 SAPs standard• Up to 5 CPs• Up to 5 specialty engines• Up to 2 zIIPs/zAAPs• 0 spares when fully configured
� M/T 2818 – Model M10– Air cooled– Single Frame– Non-raised floor option available– 30 LPARs
� Processor Units (PUs)– New processor drawer design (2
processor drawers)– 14 per system
• 2 SAPs standard• Up to 5 CPs• Up to 10 specialty engines• Up to 5 zIIPs/zAAPs• 2 dedicated spares
� When Model M10 (requires the 2nd processor drawer)?
– > 5 Customer PUs– > 120 GB memory– > 4 Fanouts for additional I/O connectivity – especially PSIFB links
• Depends - numbers vary for drawers, I/O features and PSIFB links
zEnterprise 114 Models M05 and M10
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
11
zEnterprise 114 Functions and Features (GA Driver 93 – September, 2011)
Two hardware models
Up to 10 processors configurable as CPs, zAAPs, zIIPs, IFLs, ICFs, or
optional SAPs
Up to 130 subcapacity settings across a maximum of 5 CPs
Up to 256 GB of Redundant Array of Independent Memory (RAIM) for System
Dedicated Spares on the Model M10
Increased capacity processors
Out of order instruction execution
Improved processor cache design
New and additional instructions
On Demand enhancements
CFCC Level 17 enhancements
Cryptographic enhancements
6 and 8 GBps interconnects
2 New OSA CHPIDs – OSX and OSM
Doubled HiperSockets to 32
Additional STP enhancements
Doubled Coupling CHPIDs to 128
Improved PSIFB Coupling Link
Physical Coupling Links increased to 72 (Model M10)
New 32 slot PCIe Based I/O Drawer
Increased granularity of I/O adapters
New form factor I/O adapters i.e FICON Express8S and OSA-Expres4S
Humidity and altimeter smart sensors
Optional High Voltage DC power
Optional overhead I/O cable exit
zBX-002 with ISAOPT, POWER7, DataPower XI50z and IBM System x
Blades
NRF Support with either top exit or bottom exit I/O and power
Reclassification from “general business”environment to “data center”
z114
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
12
zEnterprise zBX Functions and Features
One hardware model
zBX is controlled by one specific z196 or z114
Up to 4 Racks (B, C, D and E)
2 BladeCenters Chassis per rack
Non-acoustics doors standard
Optional Rear Door Heat Exchanger
Optional Rear acoustic door
Redundant Power, Cooling and Management Modules
10 GbE and 1000BASE-T Network modules
8 Gb FC modules
Advance Management Module
1000BASE-T and 10 GbE TORs
Up to 112 Blades
IBM Smart Analytics Optimizer
POWER7 Blades
IBM System x Blades
IBM WebSphere DataPowerIntegration Appliance XI50 for zEnterprise
HMCs required for Unified Resource Manager
Additional zBX owned HMC required if System maintained by Third Party
zBX
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
1313
Internal
Batteries
(optional)
Power
Supplies
I/O Drawer
2 x Processor Drawers, Memory
& HCAs
FICON & ESCON®
FQC
Ethernet cables for internal System LAN
connecting Flexible Service Processor
(FSP) cage controller
cards (not shown)
PCIe I/O drawers
Rear View Front View
2 x Support
Elements
z114 Model M10 – Under the covers
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
14
CEC Drawers Placement
14
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
15
One z10 BC DrawerTwo z114 Drawers (Model M10)
� System resources split between 2 drawers
(Model M10)� Second processor drawer (Model 10) for:
– Increased specialty engine capability– Increased memory capability– Increased I/O capability
•More coupling links than z10 BC•More I/O features than z10 BC
Note: Unlike the z196 Books, add/remove/repair of the processor drawer is disruptive
Processor / Memory Subsystem Drawers (Model M05 and M10)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
16
z114 Processor drawer packaging
PU SCM
SC SCM
PU SCM
HCA
HCA
HCA
HCA
inter procdrawer connect
2 x I/O Fanouts
2 x OSC
2 x FSP
2 x I/O Fanouts
2x DCA
10 x MemoryDimms
Two PU Single Chip Modules (SCMs) & One Storage Control (SC) SCM
Drw to DrwInterconnect
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
17
z114 Processor drawers Infrastructure (M10)
Processor drawer #2
Processor drawer #1
HCA2 or 3s OSC FSP HCA2 or 3s
Interconnect (in Brown) - C-shape flex cable SC to SC SMP Signals and Oscillator (OSC) Signals
DCA Power
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
18
Oscillator / Fabric Flex Cable
18
The “Oscillator / Fabric Flex Cable” which connects Drawer 0 with Drawer 1
Used to transport Oscillator signals (from Drawer 0 to Node1) and to interconnect the Fabric buses
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
19
z114 Processor drawer #1 (front) with fanout cards
2 Flexible Support Processor (FSP) card slots providing support for the Service Network subsystem
4 fanout card slots providing support for the I/O subsystem and/or coupling
2 x oscillators with Pulse Per Second (PPS)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
20
z114 Processor drawer #2 (front) with fanout cards
2 Flexible Support Processor (FSP) card slots providing support for the Service Network subsystem
4 fanout card slots providing support for the I/O subsystem and/or coupling
2 x oscillators – No PPS
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
2121
Master Oscillator Card for Model 05 with PPS connector
BNC Connector for Pulse Per Second
NTP Ethernet LAN
Note: If the NTP LAN goes through the HMC, PPS can not be used.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
22
z114 Model Structure and Upgrades
22
Model CPs
IFLs
Unassigned IFLs
zAAPs zIIPs ICFsStd.
SAPs
Add’I
SAPsSpares
M05 0-5 0-5 0-2 0-2 0-5 2 0-2 0
M10 0-5 0-10 0-5 0-5 0-10 2 0-2 2
� Model structure based on number of drawers
� M05 sparing based on prior Business Class (BC) offerings – no dedicated spares
� M10 sparing based on Enterprise Class (EC) offerings – dedicated spares
� SAP and PU Allocation/Sparing in the M10
� Default assignment is one SAP per drawer; one Spare per drawer. Spill and fill CP low to high; spill and fill specialty engines high to low
� Two defective PUs may cause the default assignment to spill and fill into the second processor drawer. LPAR has the capability to request PU of a specified type to be grouped together in a book/drawer (i.e. LPAR may change the default assignment)
� Disruptive upgrade from M05 to M10
– No model downgrades
� Upgrades from z9 BC and z10 BC into either model M05 or M10
� Only the M10 will upgrade to z196 Model M15 (Air cooled only)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
23
z114 PU Conversions
23
To
From……….. IFL
Unassigned
IFLICF zAAP zIIP
Additional SAP
IFL x Yes Yes Yes Yes Yes
Unassigned IFL Yes x Yes Yes Yes Yes
ICF Yes Yes x Yes Yes Yes
zAAP Yes Yes Yes x Yes Yes
zIIP Yes Yes Yes Yes x Yes
Additional SAP Yes Yes Yes Yes Yes x
Specialty engines are converted "any to any", including Additional SAPs and unassigned IFLs.
CP capacity conversions are handled separately from the specialty engines under the following rules:
� CP Capacities are converted "any to any"...allow MIPS upgrades, downgrades and horizontal movement.
� CP Capacities are handled separately from specialty engines, meaning that in the conditions where CP capacity on the target machine is constructed using a lesser amount of CPs, the remaining CPsare not available to be converted to specialty engines.
.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
24
z114 Conversions (continued)
24
Any to Any is allowed (upgrades or downgrades)– CP Capacities are handled separately from specialty engines.
• When CP capacity on the target machine is ordered using fewer CPs, the remaining CPs are not available to be converted to specialty engines without a fee.
F01 F02 F03 F04 F05
E01 E02 E03 E04 E05
D01 D02 D03 D04 D05
C01 C02 C03 C04 C05
B01 B02 B03 B04 B05
A01 A02 A03 A04 A05
1-way 2-way 3-way 4-way 5-way
Examples:
2-way to 1-way capacity downgrade
The PU in the green column must maintain its high water mark and can not be purchased as a specialty engine.
2-way to 1-way capacity upgrade
The high water mark is now in the red column
and the PU in the green column can be used during the purchase of a specialty engine.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
25
IBM System z Business Class Configuration Comparisons (1 of 3) z9 BC
R07
z9 BC
S07
z10 BC
E10
z114
M05
z114
M10
Uniprocessor Performance 470 MIPS 673 MIPS 782 MIPS
z/OS Capacity 26-172 MIPS 193-1748 MIPS 26-2760 MIPS 26 - 3139 MIPS
Total System Memory 64 GB 256 GB 128 GB 256 GB
Configurable Engines 7 7 10 5 10
Dedicated Spares 0 0 0 0 2
Configurable CPs 1-3 0-4 0-5 0-5
LPARS/LCSS 15/1 30/2 30/2 30/2
HiperSockets 16 16 32
I/O Cages/Drawers 1 1 Up to 4 Up to 3(1) Up to 3(1)
I/O slots per Cage/Drawers 16 28 8 8/32(2)
FICON® Channels 64 112 128 128(3)
OSA Ports (10GbE/1GbE) 16/32 24/48 48/96 48/96
ESCON® Channels 240 420 480 240(4)
STI (z9), IFB (z10) BandwidthPCIe (z114 BC) Bandwidth
2.7 GB/sec 6.0 GB/sec 6.0 GB/sec8.0 GB/sec
ICB-4/ISC-3/PSIFB 16/48/0 12/48/12 0(5)/48/8 -16(6) 0(5)/48/16 - 32(7)
zIIP/zAAP Maximum Qty 3 3 5 2 5
IFL Maximum Qty 6 7 10 5 10
Capacity Settings 20 53 130 130 130
Upgradeable Upgrade to z10 and z114 Upgrade to z114 Upgrade From M10 to z196 (M15, Air cooled only)
See next chart for foot notes
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
26
(1) Up to 3 channel drawers standard, a combination of I/O drawers and PCIe I/O drawers as defined. Up to 4 I/O drawers via RPQ
(2) 8 card slots per I/O drawer, 32 per PCIe I/O drawer
(3) FICON count is based on 2 PCIe I/O drawers or 4 I/O drawers
(4) Limit of 240 ESCON channels is consistent with Statement of Direction
(5) Limit of 0 ICB-4 links is consistent with Statements of Direction
(6) 8 ports of 12x, 16 ports of 1x PSIFB links available on model M05 based on 4 HCA capabilities
(7) 16 ports of 12x, 32 ports 1x PSIFB links available on model M10 based on 8 HCA capabilities
M05 M10
I/O Drawer PCIe I/O Drawer
I/O Drawer PCIe I/O Drawer
0 0 0 0
0 1 0 1
0 2 0 2
1 0 1 0
1 1 1 1
1 2
2 0 2 0
2 1 2 1
3 0 3 0
4 0
3rd and 4th I/O drawers are offered via an RPQ
Notes for Configuration comparisons chart (2 of 3)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
27
IBM System z Config Comparisons, z114 vs. z196 Model M15 (3 of 3)
See previous chart for foot notes
Upgrade from M10 to z196 (M15, MRU cooled only)
130
5
2
0(5)/48/8 -16(6)
6.0 GB/sec8.0 GB/sec
240(4)
48/96
128(3)
8/32(2)
Up to 3(1)
32
30/2
0 - 5
5
120 GB
26 - 3139 MIPS
782 MIPS
z114
M05
130
10
5
0(5)/48/16 - 32(7)
Up to 3(1)
10
248 GB
z1114
M10
z196 M32, M49, M66, M80MRU and Water Cooled
60
15
7
0(5)/48/16 - 32
6.0 GB/sec8.0 GB/sec
240
48/96
256
28/8/32(2)
3/6
32
60/4
0 - 15
15
704 GB
240 – 13,660 MIPS
1200 MIPS
z196
Model M15
Total System Memory
ICB-4/ISC-3/PSIFB
IFL Maximum Qty
I/O slots per Cage/Drawers
I/O Cages/Drawers
Upgradeable
Capacity Settings
zIIP/zAAP Maximum Qty
STI (z9), IFB (z10) BandwidthPCIe (z114/z196) Bandwidth
ESCON® Channels
OSA Ports (10GbE/1GbE)
FICON® Channels
HiperSockets
LPARS/CSS
Configurable CPs
Configurable Engines
z/OS Capacity
Uniprocessor Performance
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
28
IBM zEnterprise 114 (z114)
Processor Unit Design
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
29
Heatsink
PU Chip
LGA
Processor drawer with
2 PU SCMs, 1 SC SCM
Memory DIMMS and DCA Power
z114 PU/SC Single Chip Module (SCM) Components
Thermal Interface Material
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
30
z114 SCM Vs z196 MCM Comparison – Same PU and SC Chip
Single PU Chip
without heatsink
� MCM
– 96mm x 96mm in size
– 6 PU chips per MCM
• Quad core chips with 3 or 4 active cores• PU Chip size 23.498 mm x 21.797 mm
– 2 SC chips per MCM• 96 MB L4 cache per chip
• SC Chip size 24.427 mm x 19.604 mm
– Up to 4 MCMs for System
z196 Multi Chip Module (MCM)
� PU SCM
– 50mm x 50mm in size – fully assembled– Quad core chip with 3 and 4 active cores
– 2 PU SCMs for M05 and 4 PU SCMS for M10
– PU Chip size 23.498 mm x 21.797 mm
� SC SCM
– 61mm x 61mm in size – fully assembled– 1 SC SCM for M05, 2 SC SCMs for M10
– 96 MB L4 cache per chip
– SC Chip size 24.427 mm x 19.604 mm
z114 SCMs
Single SC Chip
without heatsink
PU 0PU 2
SC 0SC 1
PU 1
S00
S01
PU 5PU 3 PU 4
S10
S11
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
31
z114 PU Chip Details
� 12S0 45nm SOI Technology
– 13 layers of metal
� Chip Area – 512.3mm^2
– 23.5mm x 21.8mm
– 8093 Power C4’s
– 1134 signal C4’s
� 1.4 Billion Transistors
� 4 Core Design
– 3.8 GHz
– Full out-of-order capability
– 1.5MB L2 cache
– 12MB shared L3 cache
� Two Co-Processors
– Data Compression & encryption
� Memory controller
– Supports 5-channel DDR3 RAIM
� GX Controller
� 3 PLLs, 5+ frequency domains
� 5 unique chip voltage supplies
L2(1.5MB)
CoPMCU
L2(1.5MB)
L2(1.5MB)
CoP GX
L2(1.5MB)
L3_0 Controller
L3_1 Controller
L3B
L3B
Core 0
Core 1
Core 2
Core 3
MCIOs
MCIOs
GXIOs
GXIOs
L3 Cache(12MB)
L3 Cache(12MB)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
32
z114 SC Chip Details
�12S0 45nm SOI Technology–13 layers of metal
�Chip Area – 478.8mm^2 –24.4mm x 19.6mm–7100 Power C4’s–1819 signal C4’s
32
�1.5 Billion Transistors
–1 Billion cells for eDRAM
�eDRAM Shared L4 Cache–96 MB per SC chip
�6 CP chip interfaces
–2 used
� 3 Fabric interfaces
– 1 used
� 2 clock domains
� 5 unique chip voltage supplies
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
33
z114 System Topology
SC00
SC10
CP02
CP12
CP01
CP11
SC0 CP2 CP1
SC0
SC0
SC0CP1CP2
GX1 GX0 GX1 GX0
GX1
GX06 GX02
GX17 GX11
FP2
FP2
Drawer #1
GX0
GX07 GX01
PSI Redundant
5 DIMMs 5 DIMMs
5 DIMMs
PSI Redundant
5 DIMMs
GX1 GX0
GX16 GX12 Drawer #2
Color legend:
Same as in z196 models
Different
� 2-drawer System
� Up 14 Processing Units at 3.8 GHz
� 3 or 4 cores active, same as z196
� Up to 16 HCA ports (M10)
� Support for Legacy and PCIe HCAs
� Up to 4 Memory Controllers
� Up to 20 DIMM slots
� 5-channel Memory RAIM protection
� 4-level cache hierarchy– Split 64K I-L1 and 128K D-L1 per core– 1.5 MB L2 per core– 12 MB on chip L3 half that of z196– 96 MB node level L4 half that of z196
� eDRAM (embedded) caches – L3, L4 and SP Key Cache
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
34
z114 Internal System Structure – Compared to the z196
Book
Book
Book
Book
FBC
FBC
FBC
FBC
FBC
FBC
PU1PU2
CP4 CP5CP3
SC0SC1
GX5PSI
GX4GX3
Mem1
GX1
Mem0
GX0
FBC2
FBC0FBC1FBC2
Mem2
GX2
GX6 GX7
BOOK
CP0
Topology
FBC0FBC1
� z114 is built from subset of the z196 design and chipset– Half the L3 and L4 cache sizes– Half the chip to chip bandwidth
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
35
z114 Cache Hierarchy
35
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
3636
Cache Hierarchy – z196 (4 books) vs. z114 (2 processor drawers) Comparison
L2L1
L2L1
L2L1
L2L1
L1: 64KI + 128KD8w DL1, 4w IL1256B line size
L2 Private 1.5MB Inclusive of L1s12w Set Associative256B cache line size
L3 Shared 12MB Inclusive of L2s12w Set Associative256B cache line size
L4 96MB Inclusive24w Set Associative256B cache line size
z114
2 L4 Caches
96MB
Shared eDRAM L4
2 L3s,7 L1 / L2s
L2L1
12MB ShreDRAM L3
L2L1
L2L1
L2L1
L2L1
L2L1
L2L1
L2L1
L1: 64KI + 128KD8w DL1, 4w IL1256B line size
L2 Private 1.5MB Inclusive of L1s12w Set Associative256B cache line size
L3 Shared 24MB Inclusive of L2s12w Set Associative256B cache line size
L4 192MB Inclusive24w Set Associative256B cache line size
z196
4 L4 Caches
192MB
Shared eDRAM L4
6 L3s,24 L1 / L2s
L2L1
24MB ShreDRAM L3
L2L1
L2L1
L2L1
24MB ShreDRAM L3
12MB ShreDRAM L3
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
37
z114 Cache Structure
� On the processor chip there are 3 levels of cache
– L1 I Instruction Cache of 64KB – L1 D Data Cache of 128KB – L2 Private Cache of 1.5MB (per core) – L3 Share Cache of 12MB
� L3 Cache– 12MB eDRAM– Gear Ratio: 2:1 dataflow, 4:1 control flow – 2 address slices, address bit 55 selection – 16 octword cache interleaves – Address and Data bit stacking – Store-in cache, store dual-piped per L3 slice – 16B 2:1 store/fetch buffers planned between
L2&L3 – Extension of z10 cache coherency scheme – CP/SC merged transaction bussing
(addr/control + data) – IO fetch reserved cache slots
37
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
38
z114 Cache Structure – L4 Cache
� L4 directory is used to filter snoop traffic from the remote drawer
� Enhanced Fabric Protocol for improved latency and cache management
� Horizontal Cache Persistence for local cache extension
� LRU install on available cache slots in the remote L4 cache
� 96 MB eDRAM per SC chip, 24 Set Associatively.
� Gear ratio: 2:1 dataflow, 4:1 control flow. Synchronous fabric port
� 2 address sliced L4 Directories within the SC chip, address bit 54 selection
� 8 256B cache banks per logical directory
� z10 derived cache and storage coherency scheme
38
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
39
zEnterprise System PU core
� Each core is a superscalar, out of order processor with these characteristics:
– Six execution units
• 2 fixed point (integer), 2 load/store, 1 binary floating point, 1 decimal floating point
– Up to three instructions decoded per cycle (vs. 2 in z10)
– 211 complex instructions implemented by multiple internal operations
• 246 of the most complex z/Architecture instructions are implemented via millicode
– Up to five instructions/operations executed per cycle (vs. 2 in z10)
– Execution can occur out of (program) order
• Memory address generation and memory accesses can occur out of (program) order
• Special circuitry to make execution and memory accesses appear in order to software
– Each core has 3 private caches
• 64KB 1st level cache for instructions, 128KB 1st level cache of data
• 1.5MB L2 cache containing both instructions and data
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
40
� OOO yields significant performance benefit for compute intensive apps through
–Re-ordering instruction execution
• Later (younger) instructions can execute ahead of an older stalled instruction
–Re-ordering storage accesses and parallel storage accesses
� OOO maintains good performance growth for traditional apps
zEnterprise System Out of Order (OOO) Value
L1 miss
Instrs
1
2
3
4
5
Time
In-order core execution Out-of-order core execution
L1 miss
TimeExecution
Storage access
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
41
zEnterprise System Out of Order Detail
� Out of order yields significant performance benefit through
– Re-ordering instruction execution
• Instructions stall in a pipeline because they are waiting for results from a previous instruction or the execution resource they require is busy
• In an in-order core, this stalled instruction stalls all later instructions in the code stream
• In an out-of-order core, later instructions are allowed to execute ahead of the stalled instruction
– Re-ordering storage accesses
• Instructions which access storage can stall because they are waiting on results needed to compute storage address
• In an in-order core, later instructions are stalled
• In an out-of-order core, later storage-accessing instructions which can compute their storage address are allowed to execute
– Hiding storage access latency
• Many instructions access data from storage
• Storage accesses can miss the L1 and require 10 to 500 additional cycles to retrieve the storage data
• In an in-order core, later instructions in the code stream are stalled
• In an out-of-order core, later instructions which are not dependent on this storage data are allowed to execute
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
42
zEnterprise System Compression and Cryptography Accelerator
� Data compression engine
– Static dictionary compression and expansion
– Dictionary size up to 64KB (8K entries)
• Local 16KB cache per core for dictionary data
� CP Assist for Cryptographic Function (CPACF)
– Enhancements for new NIST standard
– Complemented prior ECB and CBC symmetric cipher modes with XTS, OFB, CTR, CFB, CMAC and CCM
– New primitives (128b Galois Field multiply) for GCM
� Accelerator unit shared by 2 cores
– Independent compression engines
– Shared cryptography engines
Core 0 Core 1
IB IBOB OBTLBTLB
2nd LevelCache
CmprExp
CmprExp
16K 16K
CryptoCipher
Crypto Hash
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
43
z196 and z114 Comparison
z114z196 Bi-NodalFully InterconnectedNodal Topology
2 drawers4 booksNumber of books / drawers
1496Max # of PU’s
2 PUs, 1 SC6 PUs, 2 SCsNumber of chips per Book/Drawer
64KB I, 128KB D64KB I, 128KB D1st level cache (PU core)
1.5MB private per core (L2)1.5MB private per core (L2)2nd level cache (PU core)
12MB Shared per 4 cores (L3)24MB Shared per 4 cores (L3)3rd level cache (PU nest)
96MB Shared per node (SC L4)192MB Shared per node (SC L4)4th level cache (SC)
1 Main, 1 Store2 Main, 2 Store# of PU L3 pipes
1 Main1 Main (per SC)# of SC cache pipes
x81 DIMMsx81 DIMMsMemory DIMMs
256 Gigabytes on 20 DIMMs3 Terabytes on 120 DIMMsMax System Memory Size
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
44
IBM zEnterprise 114 (z114)
Memory Design
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
45
z114 RAIM Memory
� Memory technology introduced on the z196 is used on the z114
– Redundant Array of Memory (RAIM) which in the Disk industry is known as RAID
– Protection from Unscheduled Incident Repair Actions (UIRAs)caused by a DIMM failure
• DIMM failures include all components on the DIMM
• Portions of the memory controller or card failure isolated to one memory channel
� Flexible memory option not available on z114
– Used for Enhanced Book Availability on z196
DATA
CHECK
DATA
CHECKECCRAIM Parity
Level 4 Cache
Key Cache
MCU 0
Key Cache
MCU 1
Extra column provides RAIM
function
16B 16B 16B 16B
2B 2B
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
46
z114 Memory Topology
� 10 DIMMSs/drawer
� No cascading
� All DIMMs in drawer must be the same
� DIMMs plugged in quints– 5 DIMMs
46
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
47
Physical base memory
• M05 = 40GB, 24GB for concurrent ordering before a physical (disruptive) card change (minus HSA/RAIM)
• M10 = 80GB, 56GB for concurrent upgrade ordering before a physical (disruptive) card change (minus HSA/RAIM)
FC1903 - Memory Capacity Increment
• Used to determine memory billing for new memory being added server configurations
• Indicates 8GB (or 32 GB in larger configurations) LICC’d increments of Memory Capacity.
• Charged by increments when Plan Ahead memory is enabled (conversion)
• Subsequent memory upgrade orders will use up the Plan Ahead memory first
FC 1904 - 16GB Feature Converted Memory
• Used to determine billing for memory on MES upgrades coming from System z9 and z10 servers..
FC1993 – Preplanned Memory
• Tracks the quantity of 8GB physical increments.
• Charged (half price) when physical memory is installed
z114 Memory Details
Model Increment (GB) Standard (GB) Plan Ahead (GB)
M05 8 8 – 120 24 – 120
M10 8 16 – 120 56 – 120
M10 32 152 - 248 152 - 248
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
48
z114 Model M05 Memory Features
� Plan Ahead Memory
– Pre-plugged memory based ontarget capacity specified bythe customer.
– Enabled by LICCC, concurrently.– FC1993 tracks the quantity of 8GB
physical increments.• Charged (half price) when physical
memory is installed– FC1903 generally indicates 8GB
(or 32 GB in larger configurations)LICC’d increments of Memory Capacity.
• Charged by increments when Plan Ahead memory is enabled• Subsequent memory upgrade orders will use up the
Plan Ahead memory first
10 x 4 GB DIMMs
10 x 8 GB DIMMs10 x 16 GB
DIMMs
Feature Size Feature Size Feature Size
8 32 64
16 40 72
24 48 80
56 88
96
104
112
120
Physical memory upgrades are DISRUPTIVE
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
49
z114 Model M10 Memory Features
4 GB/4GB4GB/8GB8GB/4GB
8GB/8GB4GB/16GB16GB/4GB
8GB/16GB16GB/8GB
16GB/16GB
Feature Size Feature Size Feature Size Feature Size Feature Size Feature Size
16 64 96 152 184 216
24 72 104 248
32 80 112
40 88 120
48
56
Physical memory upgrades are DISRUPTIVE
32 GB increments
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
50
z114 Memory Offerings
FC GB Increment M05 M10 (2 CPC drawers)
Dial Max Dimm (GB) # plugged Dial Max Dimm (GB) # plugged
3609 8 8 24 4 10 N/A N/A N/A
3610 16 8 4 10 56 4/4 10/10
3611 24 8 4 10 4/4 10/10
3612 32 8 56 8 10 4/4 10/10
3613 40 8 8 10 4/4 10/10
3614 48 8 8 10 4/4 10/10
3615 56 8 8 10 4/4 10/10
3616 64 8 120 16 10 88 4/8 10/10
3617 72 8 16 10 4/8 10/10
3618 80 8 16 10 4/8 10/10
3619 88 8 16 10 4/8 10/10
3620 96 8 16 10 120 8/8 10/10
3621 104 8 16 10 8/8 10/10
3622 112 8 16 10 8/8 10/10
3623 120 8 16 10 8/8 10/10
3624 152 32
N/A
152 4/16 10/10
3625 184 32 184 8/16 10/10
3626 216 32 248 16/16 10/10
3627 248 32 16/16 10/10
Memory upgrades within the same color (except white)are concurrent without the need for Memory Plan Ahead.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
51
System z HSA Comparisons
� HSA significantly larger than pre-z10 Servers
� Fixed HSA, does not affect customer purchased memory
� Size of HSA on prior Servers (dependant on defined configuration)
– Multiprise® 2000 From 12 MB up to 40 MB
– 9672 G4 From 48 up to 64 MB
– Multiprise 3000 From 38 MB up to 136 MB
– 9672 G5/G6 From 64 MB up to 192 MB
– z800 From 160 MB up to 256 MB
– z900 From 288 MB up to 512 MB
– z890 From 768 MB up to 1.9 GB
– z990 From 1 GB MB up to 2 GB
– z9 BC From 896 MB up to 2.7 GB
– z9 EC From 1.2 GB up to 4.2 GB
– z10 BC 8 GB – Fixed
– z10 EC 16 GB – Fixed
– z196 16 GB – Fixed
– z114 8 GB - Fixed
� HSA Estimator on Resource Link not relevant
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
52
IBM zEnterprise 114 (z114)
I/O Infrastructure
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
53
Glossary for System z I/O on z196 and z114
Acronym Full Name Description / Comments
N/A I/O drawerI/O drawer introduced with z10 BC and also supported on z196 and z114; has 8 I/O card slots
N/A I/O cageI/O cage available since z900 (not supported on z10 BC or z114);has 28 I/O card slots
N/A PCIe switchIndustry standard PCIe switch Application-specific integrated circuit(ASIC) used to fanout (or multiplex) the PCI bus tothe I/O cards within the PCIe I/O drawer
N/A PCIe I/O drawer New I/O drawer that supports PCIe bus I/O infrastructure; has 32 I/O card slots
PCI-IN PCIe interconnectCard in the PCIe I/O drawer that contains the PCIe switch ASIC z10 uses IFB-MP; z9 uses STI-MP
N/A PCIe fanout
Card on front of processor book that supports PCIe Gen2 bus; used exclusively to connect to the PCIe I/O drawer; PCIe fanout supports FICON Express8S and OSA-Express4SUsed instead of an HCA2-C fanout for I/O which continues to support the cards inthe I/O cage and I/O drawer
HCA3or
HCA3-O LR
HCA3-O LR fanoutfor 1x IFB
For 1x InfiniBand at unrepeated distances up to 10 km; 5 Gbps link data rate; 4 ports per fanout; may operate at 2.5 Gbps or 5 Gbps. Based upon capability of DWDM. Exclusive to z196 and z114; can communicate with an HCA2-O LR fanout; third generation Host Channel Adapter
HCA3or
HCA3-O
HCA3-O fanout for 12x IFB
For 12x InfiniBand at 150 meters; supports 12x IFB and 12x IFB3 protocols; increased service times when using 12x IFB3 protocol; 6 GBps link data rate; two ports per fanout; can communicate with an HCA2-O fanout on z196 or z10; cannot communicate with an HCA1-O fanout on z9; third generation Host Channel Adapter
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
54
Glossary for System z InfiniBand Coupling
Acronym Full name Comments
AID Adapter identification HCA fanout has AID instead of a PCHID
CIB Coupling using InfiniBand CHPID type z196, z114, z10, System z9
HCA Host Channel Adapter Path for communication
PSIFB Parallel Sysplex using InfiniBand InfiniBand Coupling Links
12x IFB 12x InfinBand 12 lanes of fiber in each direction
1x IFB 1x InfiniBand Long Reach - one pair of fiber
12x IFB3 12x InfiniBand3Improved service times of 12x IFB on HCA3-O
Type z10 z196/z114
HCA1-O fanout N/A (z9 only) N/A (z9 only)
HCA2-C fanoutCopper – Connects to I/O Cage (z10
EC only) or I/O DrawerCopper - Connects to I/O Cage (z196
only) or I/O Drawer
HCA2-O fanout Optical - Coupling 12x InfiniBand Optical - Coupling 12x InfiniBand
HCA2-O LR fanout Optical - Coupling 1x InfiniBand Optical - Coupling 1x InfiniBand
PCIe fanout N/A Copper - Connects to PCIe Drawer
HCA3-O fanout N/A Optical - Coupling 12x InfiniBand
HCA3-O LR fanout N/A Optical - Coupling 1x InfiniBand
MBA fanout Copper - Coupling (ICB-4) N/A
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
55
System z I/O Interface Evolution
eSTI(IBM)
InfiniBandDDR
(Industry Standard)
Extended Link(processor to IO Cage)
z9
IO Cage Backplane
mSTI(IBM)
mSTI(IBM)
ISC(IBM)
InfiniBandDDR
(Industry Standard)
Coupling Link
z10
z196
z196z114
Sep 2011
PCI Express(Ind. Std.)
PCI Express(Ind. Std.)
InfiniBandDDR
(Ind. Std.)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
56
STIz990/z890
STIz9
InfiniBandz10/z196/z114
STIz900/z800
STI: Self-Timed Interconnect
6 GBps
2.7 GBps
2 GBps
1 GBps
System z I/O Subsystem Internal Bus Interconnect Speeds (GBps)
PCIez196/z114Sep 2011
8 GBps
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
57
PCIe Internal I/O Infrastructure
� System z continues the I/O evolution with a new internal I/O infrastructure change on
the platform to continue to support the industry direction for new, high performance I/O
– Increase the internal bandwidth of I/O infrastructure backbone from InfiniBand (6 GBps) to PCIe (8 GBps)
– Allow for future I/O performance growth requirements– Provides future power management capability to reduce overall power consumption– Adopt industry standard technology (PCIe)
• System z strategic direction to utilize industry standards without compromising robustness and System z qualities of service.
• Enables improved I/O capability and flexibility by leveraging the technology investments of the industry
– Allows for the future possibility of integration of industry standard, native PCIe adapters and accelerators
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
58
PCIe I/O drawer 1 Of 2
� Compact
– Two PCIe I/O drawers occupy the same space as one I/O cage
� Increased granularity
– Two FICON Express8S channels per feature– One or two OSA-Express4S ports per feature
� I/O card density – 14% more capacity– 32 I/O card slots
� Peripheral Component Interconnect Express (PCIe Gen2) – 8 GBps bus – Infrastructure from processor book/drawer to I/O cards
� Reduced power consumption
� Designed for improved RAS
– Symmetrical, redundant cooling across all cards and power supplies– Temperature monitoring of critical Application-specific integrated circuits (ASICs) – Concurrent add/repair of PCIe I/O drawer and adapters
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
59
PCIe I/O drawer 2 of 2
� Designed to
– Support concurrent add and repair– Support for a new industry standard I/O
• Attachment of select industry standard PCIe cards– Provide a balanced system – Designed to provide improved performance for new and traditional workloads– Lower power requirements while increasing connectivity for all workloads
� Increased infrastructure bandwidth
– I/O bus infrastructure data rate up to 8GBps• Positioned for higher bandwidth in the future
� Power Save Mode– Power can be reduced on unused ports and unused area of the drawer
� Fewer ports per I/O card, but support for four times as many slots.– Intended for new I/O features (FICON Express8S, OSA-Express4S)– Legacy I/O technology (ESCON, ISC-3, PSC, Crypto Express3 and carry forward) still
supported on I/O Drawer (FC4000)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
60
z196/z114 Internal I/O Infrastructure Overview
� The z196/z114supports two different internal I/O infrastructures
– The z196 supports:• InfiniBand based infrastructure for I/O cages and I/O drawers • PCIe based infrastructure for PCIe I/O drawers with a new form factor
drawer and I/O features
– The z114 supports:• InfiniBand based infrastructure for I/O drawers • PCIe based infrastructure for PCIe I/O drawers with a new form factor
drawer and I/O features
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
61
z114 and z196 at GA2 support two different internal I/O infrastructures
� The InfiniBand I/O infrastructure first made available on z10
– InfiniBand fanouts supporting the current 6 GBps InfiniBand I/O interconnect– InfiniBand I/O card domain multiplexers with Redundant I/O interconnect in:
• The 14U, 28-slot, 7-domain I/O cage (z196 only)• The 5U, 8-slot, 2-domain IO drawer (z114 and z196)
– Selected legacy I/O feature cards• Carry forward and new build
� And a new PCI Express 2 I/O infrastructure
– PCIe fanouts supporting a new 8 GBps PCIe I/O interconnect– PCIe switches with Redundant I/O interconnect in for I/O domains in a new 7U, 32-slot,
4-domain I/O drawer (z114 and z196 GA2)– New FICON Express8S and OSA-Express4S I/O feature cards
• Based on selected industry standard PCIe I/O• Designed to:
– Reduce purchase granularity (fewer ports per card)
– Improve performance
– Increase I/O port density
– Save energy
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
62
8-slot I/O Drawer (Introduced with z10 BC)
� Supports all z10 BC and z196 GA1 I/O and
Crypto Express3 cards
� Supports 8 I/O cards, 4 front and 4 back, horizontal orientation, in two 4-card domains (shown as A and B)
� Requires two IFB-MP daughter cards, each connected to a 6 MBps InfiniBand
interconnect to activate both domains.
� To support Redundant I/O Interconnect
(RII) between the two domains, the two interconnects must be from two different
InfiniBand fanouts. (Two fanouts can support two of these drawers.)
� Concurrent add, repair.
� Requires 5 EIA Units of space
(8.75 inches ≈ 222 mm)
DCA
I/O - B
I/O - A
IFB-MP A
I/O - B
I/O - A
IFB MP B
DCA
I/O - A
I/O - B
I/O - B
I/O - A
Front
Rear
5 U
5 U
RII
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
63
7U
7U
Rear
Front� Supports only the new PCIe I/O cards
introduced with z114 and z196 GA2.
� Supports 32 PCIe I/O cards, 16 front and
16 rear, vertical orientation, in four 8-card domains (shown as 0 to 3).
� Requires four PCIe switch cards (����), each connected to an 8 MBps PCIe I/O
interconnect to activate all four domains.
� To support Redundant I/O Interconnect(RII) between front to back domain pairs 0-1 and 2-3 the two interconnects to
each pair must be from 2 different PCIefanouts. (All four domains in one of
these cages can be activated with two fanouts.)
� Concurrent field install and repair.
� Requires 7 EIA Units of space
(12.25 inches ≈ 311 mm)
Domain 0 Domain 2
� �
� �
Domain 3 Domain 1
New 32 slot PCIe I/O drawer for z114 and z196
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
64
zEnterprise I/O Drawer comparison
• 8 horizontal slots• 4 Front/4 Back• 32 FICON Ports
• 2 DOMAINS
• 32 vertical slots
• 16 Front/16 Back• 64 FICON Ports• 4 DOMAINS
PCIe I/O DrawerFC4003
I/O DrawerFC4000
(same as z10 BC)
22.75 in
8.7
5 i
n30
in
midplane
22.75 in12
.25
.75
in
30 in
midplane
I/O Cards I/O Cards
FANOUT &
FAILOVER
PROCESSOR
FANOUT &
FAILOVER
HCA
PROCESSOR
PROCESSOR DRAWER
I/O Infrastructure
PROCESSOR DRAWER
FAIL-
OVER
GX I/O Bus
EXTENDED LINKS
CAGE BUSES
GX I/O Bus
DRAWER
HCA
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
65
Top views – I/O cage, I/O drawer, PCIe I/O drawer Domains
Front – 16 Rear – 16
FSP-1, 1
PCIeinterconnect
FSP-1, 2
RII
PCIeinterconnect
PCIeinterconnect
PCIeinterconnect
I/O drawer – 8 slots2 I/O domains
I/O cage – 28 slots7 I/O domains
Front – 16 Rear – 12
16D
15C
14D
13C
12D
11C
10D
9C
8B
7A
6B
5A
4B
3A
2B
1A
28F
27E
26F
25E
24F
23E
22F
21E
20G
19G
18G
17GW
I
R
N
G
IFB-MP B
C D
G G’
E F
DCA 1
I/O-A
I/O-B
I/O-B
I/O-AI/O-A
I/O-B
DCA 2
I/O-B
I/O-A
IFB-MP B
IFB-MP A
Front - 4 Rear - 4
PCIe I/O drawer – 32 slots4 I/O domains
RIIRIIRIIRII
Redundant I/O interconnect = RII
3
3
3
3
0
0
0
0
0
0
0
0
2
2
2
2
1
1
1
1
1
1
1
1
3
3
3
3
2
2
2
2
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
66
I/O Drawer
� No changes to as used on 2098,and
2817
� 7u I/O drawer
� The addition or removal of a drawer is a concurrent MES
� IFB-MP cards as used on I/O Cage
� DCA's are unchanged from 2098 z10-BC
� Each houses 8 I/O cards over 2
domains
� Min/Max is 0/4 for new build and 0/4 for
MES
� All card assemblies in the can be added
concurrently.
66
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
67
PCIe I/O Drawer
� 7u I/O drawer
� The PCIe drawer minimum/maximum is 0/2
� The addition or removal of a PCIedrawer is a concurrent MES
� Each PCIe houses 32 I/O cards over 4 domains
� The PCIe drawer is for new PCIe based I/O only. ESCON, ISC-3 and PSC24V
cannot be plugged in a PCIe drawer
� All card assemblies in the PCIe can be
added concurrently
67
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
68
PCIe
switch
OSA-Express4S10 GbE
PCIe
switchPCIe
switch
FICON Express8S
PCIe
switch
4 GB/s PCIe
Gen2 x8
…
IFB-MP IFB-MP
I/O D
raw
er 1
I/O
Dra
we
r 1
PCIe (8x)
Memory
SC1, SC0 (FBC)
PU PU
PU PU PU
PU
PCIe (8x)
ESCON
Channels
OSA-Express31000BASE-T
Ports
2GBpsmSTI
.
FP
GA
FP
GA
FP
GA
FP
GAESCON ESCON ESCON ESCON
HCA2-C (6x))HCA2-C (6x))
Processordrawer 0
IO Drawer 2
PCIe Fanout Support•1 PCIe drawer = 2 PCIe fanouts•2 PCIe drawers = 4 PCIe fanouts
HCA2-C Fanout Support•1 I/O drawer = 2 HCA2-C•2 I/O drawer = 2 HCA2-C•3 I/O drawer = 4 HCA2-C (RPQ)•4 I/O drawer = 4 HCA2-C (RPQ)
Max of 4 fanouts per processor
drawer !
z114 Fanout Usage for I/O (Mixed PCIe and I/O drawers)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
69
PCIe
FANOUT
PCIe
FANOUT
Front
Domain 0 Domain 2
Domain 3 Domain 1
� Different PCIe Fanouts Support Domain Pairs:
– 0 and 1– 2 and 3
� Normal operation each PCIeinterconnect in a pair supports the eight I/O cards in its domain.
� Backup operation: One PCIeinterconnect supports all 16 I/O cards in the domain pair.
Rear
PCIe switch cards (����)
���� ����
���� ����
z114 M05 Redundant I/O Interconnect Example – One PCIe Drawer
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
70
z114 Fanout Usage for the I/O and PCIe drawers
Number and types of drawers Model 05 Model 10
Zero I/O drawers 0 0
One I/O drawer 2 x HCA2-C 2 x HCA2-C
One PCIe I/O drawer 2 x PCIe fanouts 2 x PCIe fanouts
Two I/O drawers 2 x HCA2-C 2 x HCA2-C
One I/O drawer + One PCIe drawer 2 HCA2-C & 2 x PCIe fanouts 2 HCA2-C & 2 x PCIe fanouts
Three I/O drawers 4 x HCA2-C 4 x HCA2-C
Two PCIe drawers 4 x PCIe fanouts 4 x PCIe fanouts
Two I/O drawers + One PCIe drawer 2 HCA2-C & 2 x PCIe fanouts 2 HCA2-C & 2 x PCIe fanouts
Four I/O drawers (RPQ) 2 x HCA2-C N/A – no drawer slots available
One I/O drawer + Two PCIe drawers N/A – not enough fanouts 2 HCA2-C & 4 x PCIe
Notes:
� Each HCA2-C and PCIe fanout has 2 ports
� Each I/O drawer has 2 I/O domains (4 I/O features per domain). All 4 domains are plugged to the HCA2-C
� Each PCIe drawer has 4 I/O domains (8 I/O features per domain). 2 domains at a time are plugged to the PCIe fanout
� 4th I/O drawer uses one of the processor drawer locations
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
71
Description F/C Ports Comments
HCA2-C copper fanout 0162 2 To I/O Drawers (FC4000)
HCA2-O 12x IB-DDR 0163 2 PSIFB Coupling (up to 150 meters)
HCA1-O 12x IB-SDR 0167 2 z9 PSIFB coupling to zEnterprise/z10
HCA2-O LR 1x IB-DDRCarry Forward only
0168 2 PSIFB Coupling – up to 10 km (unrepeated)*
PCIe copper fanout 0169 2 To PCIe I/O Drawers (FC4003)
HCA3-O LR 1x IFB 0170 4 PSIFB Coupling - up to 10 km (unrepeated)*
HCA3-O 12x IFB 0171 2 PSIFB Coupling 150 meters
HCA3-O can communicate with the z10 via HCA2-O
HCA3-O can not communicate with the z9
HCA2 on zEnterprise can communicate with z9 HCA1
*The 1X links support a distance of up to 10 km without repeaters, RPQ 8P2263 or 8P2340 is required for 20 km support.Extended distances of up to 100 km are also possible using Dense Wavelength Division Multiplexors (DWDMs), goingover 100 km requires RPQ 8P2263 or 8P2340
z114 HCA Fanout Cards – Summary
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
72
� Fanout cards – Two-port InfiniBand host channel adapters dedicated to function
– HCA2-C fanout – I/O Interconnect
• Supports FICON, ESCON, OSA, ISC-3 and Crypto Express3 cards in I/O drawer and I/O cage domains. Always plugged in pairs.
– HCA2-O fanout – 12x InfiniBand coupling links
• CHPID type – CIB for Coupling
– Fiber optic external coupling link – 150 m
– HCA2-O LR fanout – 1x InfiniBand coupling links –Long Reach (carrry forward only for z114)
• CHPID type – CIB for Coupling
• Fiber optic external coupling link – 10 km (Unrepeated), 100 km repeated
• PCIe fanout – I/O Interconnect
• Supports FICON Express8S, OSA-Express4S in PCIedrawer. Always plugged in pairs
– HCA3-O fanout – 12x InfiniBand coupling links
• CHPID type – CIB for Coupling
– Fiber optic external coupling link – 150 m
– HCA3-O LR fanout – 1x InfiniBand coupling links –Long Reach
• CHPID type – CIB for Coupling
• Fiber optic external coupling link – 10 km (Unrepeated), 100 km repeated
* Performance considerations may reduce the number of CHPIDsper port
HCA2-C IFB IFB
HCA2-O IFB IFB
Up to 16 CHPIDs – across 2 ports
HCA2-O LR IFB IFB
Up to 16 CHPIDs – across 2 ports
z196 and z114 Connectivity for Coupling and I/O - Summary
Up to 16 CHPIDs – across 2 ports*
HCA3-O IFB & IFB3 IFB & IFB3
Up to 16 CHPIDs – across 4 ports*
HCA3-O LR IFB IFB IFB IFB
PCIe fanoutPCIe PCIe
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
73
All z114 Drawer Combinations
I/O Drawers
PCIe I/ODrawers
0 1 2 3 4
0
M05 M05 M05 M05 M05
M10 M10 M10 M10
1
M05 M05 M05
M10 M10 M10
2
M05
M10 M10
z114 Drawer Combinations
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
74
IBM zEnterprise 114 (z114)
I/O Subsystem Enhancements
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
75
Three subchannel sets per LCSS (z196 only)
� Addition of a subchannel set of 64K devices to the existing two subchannel sets
� Value– Extends the amount of addressable storage for the largest System z customers– Provides a means to provide consistent device address definitions to help simplify
addressing scheme for congruous devices• Can utilize same device number in different subchannel sets
� The first subchannel set (SS 0) allow definitions of any type of device as is allowed today, (i.e. bases, aliases, secondaries, and those other than disk that do not implement the concept of associated aliases or secondaries)
� Second and third subchannel sets (SS1 and SS2) now be designated for use for disk alias devices (of both primary and secondary devices) and/or Metro Mirror secondary
devices only
� CHPID types: ESCON - CNC, FICON - FC (both native FICON and zHPF paths)
� Features: ESCON, FICON Express8, FICON Express4
� z/OS V1.12
� z/OS V1.10 and V1.11 with PTFs
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
76
IPL for alternate subchannel set
� Allows to IPL from subchannel set 1 (SS1) or subchannel set 2 (SS2), in addition to
subchannel set 0.
� Devices used early during IPL processing can now be accessed using subchannel set
1 or subchannel set 2. This is intended to allow the users of Metro Mirror (PPRC) secondary devices defined using the same device number and a new device type in an alternate subchannel set to be used for IPL, IODF, and standalone dump volumes
when needed.
� IPL from an alternate subchannel set is supported by z/OS V1.13, as well as V1.12 and
V1.11 with PTFs and applies to the FICON and zHPF protocols.
Note: SS2 for z196 only
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
77
IBM zEnterprise 114 (z114)
I/O Connectivity Features
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
78
z114 I/O Features supported
I/O cage and I/O drawer
– Crypto Express3 (1P for z114 only)
– ESCON (240 or fewer)
– FICON Express8 (Carry forward or RPQ)
– FICON Express4 (Carry forward only for 4 port cards)
– FICON Express4-2C (Carry forward or RPQ – z114)
– ISC-3
– OSA-Express3 1000BASE-T (Includes 2P for z114)
– OSA-Express3 (Carry forward or RPQ)
• 10 GbE LR and SR, GbE LX and SX (2P for z114 only)
– OSA-Express2 (Carry forward only)
• GbE LX and SX, 1000BASE-T
– PSC (Carry forward or new build, no MES add)
PCIe I/O drawer
– FICON Express8S
• SX and 10 km LX
– OSA-Express4S
• 10 GbE LR and SR
• GbE SX and LX
8 slot I/O drawer
32 slot PCIe I/O drawer
Supported features
Note: PSIFB links are direct from the HCAs and not from the I/O cage or drawers
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
79
� I/O Channels
FICON (before FICON Express4)
FCV – ESCD Model 5 Bridge Card
� Networking
OSA-Express2 10 GbE LR
OSA-Express (pre OSA-Express2)
� Coupling Links
ICB-4 and earlier ICB
� Crypto
Crypto Express2 and earlier
� ETR
Sysplex Timer® (ETR) Attachment
Non-Supported Channel Types
z196 Non-Supported Features
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
80
FICON Express8S – SX and 10KM LX in the PCIe I/O drawer
� For FICON, zHPF, and FCP environments– CHPID types: FC and FCP
• 2 PCHIDs/CHPIDs
� Auto-negotiates to 2, 4, or 8 Gbps
� Increased performance compared to FICON Express8
� 10KM LX - 9 micron single mode fiber– Unrepeated distance - 10 kilometers (6.2
miles)– Receiving device must also be LX
� SX - 50 or 62.5 micron multimode fiber– Distance variable with link data rate and
fiber type– Receiving device must also be SX
� 2 channels of LX or SX (no mix)
� Small form factor pluggable (SFP) optics– Concurrent repair/replace action for each SFP
LX SX
FC 0409 – 10KM LX, FC 0410 – SX
OR
2, 4, 8 Gbps
2, 4, 8 Gbps
FLASH
SFP+
SFP+IBM ASIC
IBM ASIC
PCIeSwitch
HBAASIC
HBAASIC
FLASH
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
81
FICON Express8 in the I/O Drawer
� Auto-negotiate to 2, 4, or 8 Gbps
� Connector – LC Duplex
� 10KM LX – 9 micron single mode fiber
– Unrepeated distance - 10 kilometers (6.2 miles)
– Receiving device must also be LX
� SX – 50 or 62.5 micron multimode fiber
– Variable with link data rate and fiber type
– Receiving device must also be SX
� 4 channels of LX or SX (no mix)
LX SX
FC 3325 – 10KM LX, FC 3326 – SX
OR
PCIe
PCIe
PCIe
PCIe2, 4, 8 Gbps
2, 4, 8 Gbps
2, 4, 8 Gbps
2, 4, 8 Gbps
Small Form Factor Pluggable (SFP) optics. Concurrent repair/replace action for each SFP
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
82
2, 4, 8Gbps
2, 4, 8Gbps
2, 4, 8Gbps
2, 4, 8Gbps
FC 332510KM LX
LC Duplex
FICON Express8 10KM LX in the I/O Drawer
� FICON Express8
– Can be shared among LPARs, and defined as spanned
– High Performance FICON for System z (zHPF)
� 8 Gbps with Auto-negotiate capability (2, 4, or 8 Gbps)
– 1 Gbps NOT supported
� LX – 9 micron single mode fiber
– Unrepeated distance - 10 kilometers (6.2 miles)
– Receiving device must also be LX
� Small Form Factor pluggable optics for concurrent repair/replace
� Personalize as:
– FC
• Native FICON
• Channel-To-Channel (CTC)– z/OS, z/VM, z/VSE, z/TPF, TPF, Linux on System z
– FCP (Fibre Channel Protocol)
• Support of SCSI devices– z/VM, z/VSE, Linux on System z
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
83
FC 3326 SX
LC Duplex
FICON Express8 SX in the I/O Drawer
� FICON Express8
– Can be shared among LPARs, and defined as spanned
– High Performance FICON for System z (zHPF)
� 8 Gbps with Auto-negotiate capability (2, 4, or 8 Gbps)
– 1 Gbps NOT supported
� SX - 50 or 62.5 micron multimode fiber
– Unrepeated distance (20 to 500 meters) varies with speed and fiber type
– Receiving device must also be SX
� Small Form Factor pluggable optics for concurrent repair/replace
� Personalize as:
– FC
• Native FICON
• Channel-To-Channel (CTC)
–z/OS, z/VM, z/VSE, z/TPF, TPF, Linux on System z
– FCP (Fibre Channel Protocol)
• Support of SCSI devices
–z/VM, z/VSE, Linux on System z
2, 4, 8Gbps
2, 4, 8Gbps
2, 4, 8Gbps
2, 4, 8Gbps
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
84
zEnterprise zHPF supports data transfers larger than 64 k bytes
� zHPF multi-track data transfers are no longer limited to 64 k bytes– Up to 256 tracks can be transferred a single operation– Eliminating the 64 k byte limit is designed to allow a FICON Express8
channel to fully exploit its available bandwidth– This enhancement is exclusive to z196 and z114
� Designed to help provide
– Higher throughput for zHPF multi-track operations– With lower response time
� Requires: – FICON Express8 and 8S or FICON Express4 channel– CHPID TYPE=FC definition– Control unit support for zHPF
� z/OS operating system support
White Paper: “High Performance FICON (zHPF) for System z Analysis”http://www.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP101789
High Performance FICON (zHPF) for DS8000 System z Attached Analysis: AG Storage ATS Offeringhttp://www.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/FLASH10668
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
85
FICON performance on System z
1200
14000
31000
20000
52000
23000
92000
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
I/O driver benchmark I/Os per second4k block sizeChannel 100% utilized
FICONExpress4
andFICON
Express2
zHPF
FICON Express8
zHPF
FICON Express8FICON
Express4and
FICONExpress2
ESCON
zHPF
FICON Express8S
FICON Express8S
z10 z10z196z10
z196z10
z196z114
z196z114
350
520
620
770
620
1600
0
100
200
300
400
500
600
700
800
900
1000
1100
1200
1300
1400
1500
1600
1700
FICON Express44 Gbps
I/O driver benchmarkMegaBytes per secondFull-duplexLarge sequentialread/write mix
FICONExpress44 Gbps
FICON Express88 Gbps
FICON Express88 Gbps FICON
Express8S8 Gbps
FICON Express8S
8 Gbps
z10z9 z10
z196z10
z196z10
z196z114
z196z114
zHPF
zHPF
zHPF
77% increase
108% increase
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
86
FCP performance on System z
15750
31500
60000
84000 84000
92000
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
FE8 FE8 FE8S
8 Gbps 8 Gbps 8 Gbps
I/Os per secondRead/writes/mix 4k block size, channel 100% utilized
z9 z9 z10 z10 z196
z196z114
520
770
1600
0
200
400
600
800
1000
1200
1400
1600 MegaBytes per second (full-duplex)Large sequential Read/write mix
z10z196z10
z196z114
108% increase
10% increase
FE44 Gbps
FE44 Gbps
FE44 Gbps
FE8S8 Gbps
FE88 Gbps
FE44 Gbps
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
87
Buffer credits for FICON Express8 and 8S
� 40 buffer credits available per channel
� Conforms to Fibre Channel standard flow control– Avoids overwriting the buffer– 40 credits is 80 KB of buffer at the receiver – Sufficient to hold the maximum data payload on the fiber for 20 km (round trip) of data. Anything less
will not fill the buffer
� Receiver side of port should contain sufficient buffer credits to ensure channel is fully utilized
– Receiver must contain = > 4 buffer credits per km of fiber distance at 8 Gbps
� Buffer credits are directly related to the distance supported– 8 Gbps needs about 4 credits/km... – 10 km = about 40 credits. 100 km = about 400 credits– 4 Gbps needs about 2 credits/km...
• 10 km = about 20 credits. 100 km = about 200 credits
� In the short block case, the concern is response time, not data bandwidth. Where the buffer-to-buffer credits (BtoB credits) becomes an issue is when large blocks per I/O
are being transferred, so link bandwidth becomes the factor of interest. In this case, the frames will be almost always full (2K).
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
88
z/OS Discovery and Auto-Configuration (zDAC) for z196/z114
� Provides automatic discovery for FICON disk and tape control units
� Reduces level of IT skill and time required to configure new I/O devices
� Ensures system (host) and control unit definitions are compatible with each other
� Automatically discovers storage devices accessible to the system but not currently configured and proposes host definition values
– For those discovered control units, explore for defined logical control units and devices – Compare discovered logical control units and devices against those configured previously– Add missing logical control units and devices to the configuration, proposing control unit and device
numbers, and proposing paths to reach them• Channel paths chosen using algorithm to minimize single points of failure
� Integrated into existing System z host configuration tools (HCD & HCM)
� Requirements:– z114 or z196 server with FICON Express8S, FICON Express8 or FICON Express4 – Switch/director attached fabric (no direct attachment)– z/OS V1.12 (at least 1 LPAR for Dynamic I/O capability)– First exploiter is IBM System Storage DS8700
• DS8000 licensed machine code level 6.5.15.xx (bundle version 75.15.xx.xx), or later• see IBM PSP buckets
– Suggested: FICON DCM (z/OS dynamic channel management) to help manage performance
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
89
zDAC Value
Without I/O discovery and auto configuration, using HCD **
With I/O discovery and autoconfiguration**
Defining the fibre channel network
Control unit address and ports specified on paper; requires many people and transcribing from paper to HCD.
Steps: multiple iterative steps
HCD: several panels per control unit
Time: hours
Discover the connections in the fiber channel network.
HCD invokes IOS to discover I/O hardware in the current configuration based on autoconfiguration policies.
Steps: 1
HCD: 5 panels for whole fibre channel network; policy definition and running auto discovery.
Time: a few minutes
Defining the control units: building the I/O configurations
Control Unit addresses and ports are specified manually one at a time with ‘new’ or ‘’new based on’ function in HCD.
Steps: multiple iterative steps
HCD: 10 panels per control unit
Time: hours
From the discovered control units, HCD retrieves and proposes control unit and device types and numbers, channel path assignments, partition access and OS device parameters.
Definitions are automatically written into a specified target work IODF which is created as a copy of the active or accessed IODF.
Steps: 1
HCD: 5 panels for whole fibre channel network
Error checking Errors not found until IODF is activated or device is brought online, requiring rework.
Manual error checking
Addresses and ports discovered and defined electronically, using best practices ad existing configurations.
Errors discovered in working IODF and can be corrected before activation
** Based on IBM laboratory results, your results may vary
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
9090
FCP channels to support T10-DIF for enhanced reliability
� System z Fibre Channel Protocol (FCP) has implemented support of the American
National Standards Institute's (ANSI) T10 Data Integrity Field (DIF) standard. – Data integrity protection fields are generated by the operating system and
propagated through the storage area network (SAN). – System z helps to provide added end-to-end data protection between the operating
system and the storage device
� An extension to the standard, Data Integrity Extensions (DIX), provides checksum
protection from the application layer through the host bus adapter (HBA), where cyclical redundancy checking (CRC) protection is implemented
� T10-DIF support by the FICON Express8S and FICON Express8 features, when defined as CHPID type FCP, is exclusive to z196 and z114.
� Exploitation of the T10-DIF standard requires support by the operating system and the storage device
– z/VM 5.4 with PTFs for guest exploitation– Linux on System z distributions:
• IBM is working with its Linux distribution partners to include support in future Linux on System z distribution releases. .
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
91
Storage Connectivity Options
Description F/C Ports Available Comments
FICON Express8S 10KM LX 0409 2 New Initial orders
FICON Express8S SX 0410 2 New Initial orders
FICON Express4-2C SX 3318 2 Carry Forward z114 Only
FICON Express4 10KM LX 3321 4 Carry Forward
FICON Express4 SX 3322 4 Carry Forward
FICON Express4-2C 4KM LX 3323 2 Carry Forward z114 Only
FICON Express4 4KM LX 3324 4 Carry Forward
FICON Express8 10KM LX 3325 2 NewOrder with RPQ 8P2534 if FC4000 slots are open
FICON Express8 SX 3326 2 NewOrder with RPQ 8P2534 if FC4000 slots are open
Maximum FICON features varies with mix of Drawers types and Model of the System
All use LC Duplex connectors
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
92
OM1 62.5 micron at 200 MHz-km, OM2 50 micron at 500 MHz-km, OM3 50 micron at 2000 MHz-kmInter-Switch Links (ISLs) is the link between two FICON directors; FICON features do not operate at 10 Gbps* The link loss budget is the channel insertion loss as defined by the standard. # This distance and dB budget applies to FICON Express4 4KM LX features
1 Gbps 2 Gbps 4 Gbps 8 Gbps 10 Gbps ISLs
FiberCore (µ )
Light source
Distancemeters
feet
* Linkloss
budget
Distancemeters
feet
* Linkloss
budget
Distancemeters
feet
* Linkloss
budget
Distancemeters
feet
* Linkloss
budget
Distancemeters
feet
* Linkloss
budget
9µ SMLX laser
10 km6.2
miles7.8 dB
10 km6.2 miles
7.8 dB10 km
6.2 miles7.8 dB
10 km6.2 miles
6.4 dB10 km
6.2 miles6.0 dB
9µ SMLX laser
4 km #2.5
miles4.8 dB #
4 km #2.5 miles
4.8 dB #4 km #
2.5 miles4.8 dB # N/A N/A N/A N/A
50µ MM OM3SX laser
860 m2822 ft
4.62 dB500 m1640 ft
3.31 dB380 m1247 ft
2.88 dB150 m492 ft
2.04 dB300 m984 ft
2.6 dB
50µ MM OM2SX laser
500 m1640 ft
3.85 dB300 m984 ft
2.62 dB150 m492 ft
2.06dB50 m164 ft
1.68 dB82 m269 ft
2.3 dB
62.5µ MM OM1SX laser
300 m984 ft
3.00 dB150 m492 ft
2.10 dB70 m230 ft
1.78 dB21 m69 ft
1.58 dB33 m108 ft
2.4 dB
Fibre Channel Physical Interface (FC-PI-4) standard (Rev. 8.00)
� Applies to FICON Express8 and 8S (2, 4, 8 Gbps), FICON Express4 (1, 2, 4 Gbps) FICON Express2 (1, 2 Gbps), and FICON Express (1, 2 Gbps) features
� CHPID types FC (FICON, zHPF, CTC) and FCP (Fibre Channel Protocol)
� Unrepeated distances in kilometers (km), meters (m), and feet (ft)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
93
IBM zEnterprise 114 (z114)
Network connectivity
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
94
The above statements are based on OSA-Express3 performance measurements performed in a test environment on a System z10 EC and do not represent actual field measurements. Results may vary.
System z OSA-Express3 and OSA-Express4S
� Double density of ports compared to OSA-Express2
– Reduced CHPIDs to manage
– Reduced I/O slots
– Reduced I/O cages or I/O drawers
– Up to 96 LAN ports versus 48
� New microprocessor and hardware data router – compared to OSA-Express2– Large send for IPv4 and IPv6 traffic– Checksum offload– Large send packet construction, inspection and routing performed in hardware
� Designed to reduce the minimum round-trip networking time between z196/z114 systems (reduced latency)
– Designed to improve round trip at the TCP/IP application layer• OSA-Express3 and 4S10 GbE
45% improvement compared to the OSA-Express2 10 GbE
• OSA-Express3 and 4S GbE
45% improvement compared to the OSA-Express2 GbE
– Designed to improve throughput (mixed inbound/outbound)
• OSA-Express3 and 4S 10 GbE1.0 GBytes/ps @ 1492 MTU1.1 GBytes/ps @ 8992 MTU3-4 times the throughput of OSA-Express2 10 GbE0.90 of Ethernet line speed sending outbound 1506-byte frames1.25 of Ethernet line speed sending outbound 4048-byte frames
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
95
OSA-Express4S GbE and 10 GbE fiber for the PCIe I/O drawer
� 10 Gigabit Ethernet (10 GbE)
– CHPID types: OSD, OSX– Single mode (LR) or multimode (SR)
fiber– One port of LR or one port of SR
• 1 PCHID/CHPID
� Gigabit Ethernet (GbE)
– CHPID types: OSD (CHPID OSN not supported)
– Single mode (LX) or multimode (SX) fiber
– Two ports of LX or two ports of SX• 1 PCHID/CHPID
� Small form factor optics – LC Duplex
FC 0406 – 10 GbE LR, FC 0407 – 10 GbE SR
PCIeIBM ASIC
FPGA
PCIeIBM ASIC
FPGA
FC 0404 – GbE LX, FC 0405 – GbE SX
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
9696
Open Systems Adapter performance
� OSA processor becomes more efficient as throughput increases
� Window size
– TCP window determines amount of data that the sender can transmit to receiver without needing an acknowledgment from the receiver
– Faster and longer networks require larger windows to keep data flowing smoothly
� Blocking
– Performance is affected by the amount of data blocked together for transfer between OSA and TCP
� Frame size– Larger frames perform better– Larger frames reduce host and OSA processing costs– Size of frame depends on LAN type, MTU setting, size of data sent
� Measure: throughput, transaction response time, server utilization– Bulk data transfer and interactive transactions
� QDIO and jumbo frames (8992 byte MTUs) yield the highest streams
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
9797
OSA-Express4S 10 GbE performance (laboratory)Mixed Streams – 1492 Byte MTUs
1240
2080
0
500
1000
1500
2000
2500
MB
ps
680
1180
0
200
400
600
800
1000
1200
1400
MB
ps
615
1120
0
200
400
600
800
1000
1200
1400
MB
ps
11801680
0
500
1000
1500
2000
2500
MB
ps
Inbound Streams – 1492 Byte MTUs
OSA-E3 OSA-E4S
80% increase
70% increase 70% increase
40% increase
Mixed Streams – 8000 Byte MTUsInbound Streams – 8000 Byte MTUsOSA-E3 OSA-E4S
OSA-E3 OSA-E4S OSA-E3 OSA-E4S
Notes:� AWM on z/OS� z/OS is doing checksum� 1 megabyte per second(MBps) is 1,048,576bytes per second
MBps represents payloadthroughput (doesnot count packetand frame headers)
Maximum transmission units (MTUs)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
98
OSA-Express4S - CHPID type controls operation
CHPID type Purpose / Traffic Operating Systems
OSDAll OSA features
z196, z114 z10, z9, zSeries
Queue Direct Input/Output (QDIO) architectureTCP/IP traffic when Layer 3 (uses IP address)
Protocol-independent when Layer 2 (uses MAC address)
z/OS, z/VMz/VSE, z/TPF
Linux on System z*
OSX10 GbE
z196, z114
OSA-Express for zBX. Connectivity and access control to intraensemble data network
(IEDN) from z196 or z114 to zBXz/OS, z/VM
*IBM is working with its Linux distribution partners to include support in future Linux on System z distribution releases.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
99
PCIe
PCIe
OSA-Express3 – GbE and 10 GbE ‘fiber’ for the I/O Drawer
� Double the port density of OSA-Express2
� Reduced latency & improved throughput
– Ethernet hardware data router
� Improved throughput – standard & jumbo frames– New microprocessor– New PCIe adapter
� CHPID types– 10 Gigabit Ethernet – OSD, OSX– Gigabit Ethernet – OSD and OSN
• CHPID type OSN is for LPAR-to-LPAR communication. Does not use ports
CHPID shared by two ports4 ports - FC 3362 GbE LX, FC 3363 GbE SX
2 ports FC 3373 GbE SX for z10 BC and z114 only
FC 3370 10 GbE LR FC 3371 10 GbE SR
PCIe
PCIeOSA-Express2 OSA-Express3
Microprocessor500 MHz – 10 GbE448 MHz – 1 GbE
667 MHz
PCI bus PCI-X PCIe G1
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
100
OSA-Express3 1000BASE-T ‘copper’ for I/O Drawer
� Auto-negotiation to 10, 100, 1000 Mbps
� Double the port density of OSA-Express2
� Reduced latency & improved throughput
– Ethernet hardware data router
� Improved throughput – standard & jumbo frames
– New microprocessor– New PCIe adapter
� Usage of ports– OSC, OSD, OSE can exploit four ports– OSM, Port 0 only– CHPID type OSN is for LPAR-to-LPAR
communication. Does not use ports
OSA-Express2 OSA-Express3
Microprocessor 448 MHz 667 MHz
PCI bus PCI-X PCIe G1
CHPID shared by two ports4 ports - FC 3367 1000BASE-T
2 ports - FC 3369 for z10 BC and z114 only
PCIe
PCIe
Mode CHPID Description
OSA-ICC OSC TN3270E, non-SNA DFT, IPL CPCs, and LPARs, OS system console operations
QDIO OSD TCP/IP traffic when Layer 3, Protocol-independent when Layer 2
Non-QDIO OSE TCP/IP and/or SNA/APPN/HPR traffic
OSA for NCP (LP-to-LP)
OSN NCPs running under IBM Communication Controller for Linux (CDLC)
Unified Resource Manager
OSM Connectivity to intranode management network (INMN) from z196/z114 to zBX
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
101
OSA solutions – how do they measure up?
� Want to compare?
� Use IBM Application Workload Modeler (AWM)– Replaces Netmarks, and is now externally available– For z/OS and Linux on System z
� Can model your current network configuration in your environment
� Can help you allocate existing system resources more efficiently by modeling, generating real traffic on your network, and evaluating the network performance of
existing workloads
� Rerun the same workload against a proposed alternative (e.g. OSA gigabit)
� Access information at– http://www.ibm.com/software/network/awm/index.html
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
102
� Two new OSA CHPID types to support new types of z196/z114 networks� A z196/z114 System can have up to 6 types of OSA CHPID’s
– External (customer managed) networks• Defined as OSC,OSD,OSE, & OSN • Existing customer provided and managed OSA ports used for access to the current customer external
networks - no changes– Intranode management network (INMN)
• Defined as CHPID type OSM, OSA-Express for Unified Resource Manager • When the PCIe adaptor on 1000BASE-T is defined as CHPID type OSM, the second port cannot
be used for anything else • OSA-Express3 1000BASE-T configured as CHPID type OSM for connectivity to INMN from z196/z114
to Unified Resource Manager functions• OSA connection via the Bulk Power Hub (BPH) on the z196/z114 to the Top of the Rack (TORs)
switches on zBX– Intraensemble data networks (IEDN)
• Defined as CHPID OSX, OSA-Express for zBX• OSA-Express3 or 4S 10 GbE configured as CHPID type OSX for connectivity and access control to
IEDN from z196/z114 to zBX
� Functions Supported:– Dynamic I/O support– HCD– CP Query capabilities– Ensemble Management for these new channel paths and their related subchannels.
� z/VM 5.4 – CHPID types OSX and OSM cannot be varied online
Open Systems Adapter CHPIDs types:OSM (Express3 1000BASE-T)) and OSX (Express3 and 4S 10 GbE)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
103
CHPID Types OSX and OSM – Usage
OSM (INMN)
Supports IOCP CHPID types:OSC, OSD, OSE, OSN, and OSM (ONLY 1000BASE-T).
PCHID = xx0 & xx1
OSA-Express3 10 GbE2 CHPIDs1 PORT/CHPID
OSX (IEDN)
Supports IOCP CHPID types:
OSD and OSX (ONLY 10 GbE).PCHID = xx0 & xx1
OSX IOCDS EXAMPLE:•CHPID PCHID=5E1,PATH=(CSS(0,1,2,3),2F),TYPE=OSX,SHARED, ….•CNTLUNIT CUNUMBR=09F0,PATH=((CSS(0),2F)),UNIT=OSX • IODEVICE ADDRESS=(09F0,15),CUNUMBR=(09F0),UNIT=OSA,UNITADD=00, *
MODEL=X,DYNAMIC=YES,LOCANY=YES
OSM IOCDS EXAMPLE:• CHPID PCHID=191,PATH=(CSS(0,1,2,3),23),TYPE=OSM,,SHARED,• CNTLUNIT CUNUMBR=0910,PATH=((CSS(0),23)),UNIT=OSM• IODEVICE ADDRESS=(0910,15),CUNUMBR=(0910),UNIT=OSA,UNITADD=00, *
MODEL=M,DYNAMIC=YES,LOCANY=YES
OSA-Express31GbE 2 CHPIDs2 PORTS/CHPID
FC3367
CAT 6 ETH CABLE
IEDN Distances•MM (Short Range Optics)50 micron at 2000 MHz-km: 300 meters (984’) 50 micron at 500 MHz-km: 82 meters (269’) 62.5 micron at 200 MHz-km: 33 meters (108’)
•SM (Long Range Optics) 10 km (6.2 miles)
MUST USE CHPID PORT 0
CHPID PORT 0
CHPID PORT 0
OSA-Express4S 10 GbE1 CHPID1 PORT/CHPID
FC3370 and FC0406 (LR)SM 9 micron LC duplex
FC3371 and FC0407 (SR)MM 50/62.5 micron LC duplex
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
104
z196/z114 HiperSockets – doubled the number
HiperSockets
L3
z/VM
LP4
L2
z/VM
LP5
z/VSE
LP3
Linux
LP2z/OS
LP1
� High-speed “intraserver” network
� Independent, integrated, virtual LANs
� Communication path – system memory
� Communication across LPARs– Single LPAR - connect up to 32 HiperSockets
� Support for multiple LCSS's & spanned channels
� Virtual LAN (IEEE 802.1q) support
� HiperSockets Network Concentrator
� Broadcast support for IPv4 packets
� IPv6
� HiperSockets Network Traffic Analyzer (HS NTA)
� No physical cabling or external connections required
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
105105
Inbound Workload Queuing for Enterprise Extender
� For improved scalability and performance
� Differentiate and separate inbound EE traffic to a new input queue
� Each input queue represents a unique type of workload
– Unique service and processing requirements– z/OS makes decision on processing resources required
• Avoid resource contention• “Off the wire” network traffic separation
� OSA-Express4S and OSA-Express3 (CHPID types OSD and OSX)
� z196 and z114
� z/OS V1.13.
� z/VM V5.4 with PTFs for guest exploitation
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
106106
CPU CPU CPU CPU
Other Bulk SD EE
Q1 Q2 Q3 Q4
z/OS Instance
(TCP/IP stack)
OSA-Express
Device
WAN
IWQ
EE traffic separation optimizes the host inbound EE processing
Inbound Workload Queuing for Enterprise Extender
� OSA separates inbound network traffic to
multiple input queues (based on type of workload) on the same host (device) interface
� Each input queue can be serviced concurrently by separate processors
� Stack receives pre-sorted packets and host processing is optimized based on the unique requirements for each unique type of workload
� New – Enterprise Extender traffic is added as a
new input queue (new IWQ)!
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
107107
� Checksum offload for IPv6 packets is now available for z/OS environments: – When the checksum function is offloaded from the host, CPU cycles are reduced, improving
performance. With the introduction of OSA-Express4S, the checksum offload function is now performed for IPv6 packets as well as IPv4 packets, whether the traffic goes out to the local area network (LAN), comes in from the LAN, or flows logical partition-to-logical partition through OSA-Express4S.
– Checksum offload provides the capability of calculating the Transmission Control Protocol (TCP), User Datagram Protocol (UDP), and Internet Protocol (IP) header checksums for Internet Protocol Version 4 (IPv4) packets and now, IPv6 packets. Checksum verifies the correctness of files.
– When checksum offload was introduced in May of 2003, it was limited to IPv4 packets. Checksum offload for IPv6 packets is exclusive to OSA-Express4S features (CHPID types OSD and OSX) on the z196 and z114. It is supported by z/OS. Refer to the Software requirements section. Checksum offload for IPv4 packets is currently available for all in-service releases of z/OS and Linux on System z.
– Checksum offload for LPAR-to-LPAR traffic in the z/OS environment is included in the OSA-Express4S design for both IPv4 and IPv6 packets.
� Large send for IPv6 packets is now available for z/OS environments: – Large send (also referred to as TCP segmentation offload) is designed to improve performance by
offloading outbound TCP segmentation processing from the host to an OSA-Express4S feature by employing a more efficient memory transfer into OSA-Express4S.
Note: Large Send for IPv6 packets is not supported for LPAR-to-LPAR packets.
� Large send support for IPv6 packets applies to the OSA-Express4S features (CHPID type OSD and OSX), and is exclusive to the z196 and z114. It is supported by z/OS and z/VM for guest exploitation
� Large send for IPv4 packets is currently available for all in-service releases of z/OS and Linux on System z.
IPv6 Support
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
108
Open Systems Adapter in the PCIe I/O Drawer (FC4003)
Description F/C Ports Available CHPID
OSA-Express4S GbE LX 0404 2 New Build OSD
OSA-Express4S GbE SX 0405 2 New Build OSD
OSA-Express4S 10 GbE LR 0406 1 New Build OSD, OSX
OSA-Express4S 10 GbE SR 0407 1 New Build OSD, OSX
Note: OSA-Express3 1000BASE-T requires an I/O Drawer (FC4000)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
109
Open Systems Adapter in the I/O Drawer (FC4000)
Description F/C Ports Available CHPID
OSA-Express 23xx/13xx 2 NO
OSA-Express2 GbE LX 3364 2 Carry Forward OSD, OSN
OSA-Express2 GbE SX 3365 2 Carry Forward OSD, OSN
OSA-Express2 1000BASE-T 3366 2 Carry Forward OSD, OSE, OSC, OSN
OSA-Express2 10 GbE LR 3368 1 NO
OSA-Express3 GbE LX 3362 41 Carry Forward OSD, OSN
OSA-Express3 GbE SX 3363 41 Carry Forward OSD, OSN
OSA-Express3 1000BASE-T 3367 41 New/Carry Forward OSD, OSE, OSC, OSN, OSM
OSA-Express3-2P 1000BASE-T 3369 21 New/Carry Forward OSD, OSE, OSC, OSN
OSA-Express3 10 GbE LR 3370 2 Carry Forward OSD, OSX
OSA-Express3 10 GbE SR 3371 2 Carry Forward OSD, OSX
OSA-Express3-2P GbE SX 3373 21 Carry Forward OSD, OSN
OSA-Express3 cards will be shown as selections via 8P2534 only if FC4000 slots are open.
1 two ports per CHPID
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
110
Summary: OSA-Express CHPID types to control operation
CHPID type Purpose / Traffic Operating Systems
OSDAll OSA features
z196, z114, z10, z9, zSeries
Supports Queue Direct Input/Output (QDIO) architectureTCP/IP traffic when Layer 3 (uses IP address)
Protocol-independent when Layer 2 (uses MAC address)
z/OS, z/VMz/VSE, z/TPF
Linux on System z
OSE1000BASE-T
z196, z114, z10, z9, zSeries
Non-QDIO; for SNA/APPN/HPR trafficand TCP/IP “passthru” traffic
z/OS, z/VMz/VSE
OSC1000BASE-T
z196, z114, z10, z9, z990, z890
OSA-Integrated Console Controller (OSA-ICC)Supports TN3270E, non-SNA DFT to IPL CPCs & LPs
z/OS, z/VMz/VSE
OSM1000BASE-T
z196 and z114
OSA-Express for Unified Resource ManagerConnectivity to intranode management network (INMN)
from z196 or z114 to Unified Resource Manager functions
z/OS, z/VMLinux on System z
OSNGbE, 1000BASE-Tz196, z114, z10, z9
No OSN support for OSA-Express4S GbE
OSA-Express for NCPAppears to OS as a device supporting CDLC protocol
Enables Network Control Program (NCP) channel-related functionsProvides LP-to-LP connectivity
OS to IBM Communication Controller for Linux (CCL)
z/OS, z/VMz/VSE, z/TPF
Linux on System z
OSX10 GbE
z196 and z114
OSA-Express for zBXConnectivity and access control to intraensemble data network (IEDN)
from z196 or z114 to zBX
z/OS, z/VMLinux on System z
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
111
IBM zEnterprise 114 (z114)
Coupling Links Connectivity and STP
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
112
ISC-3 coupling links ISC-3 coupling links
� InterSystem Channel-3 (ISC-3)
– ISC-3 links ordered in increments of one– Activated links balanced across features
� Peer mode only – 2 Gbps– FC 0217 (ISC-M), FC 0218 (ISC-D / ISC link)– Activate link – FC 0219– Four links per ISC-M
• Two links per ISC-D
– Supports 9µ single mode fiber
� Up to 48 links per machine
� Distances:– Maximum distance is 10 km, RPQ 8P2197 is required for 20 km support– Extended distances of up to 100 km are possible using Dense Wavelength Division
Multiplexors (DWDMs)– Over 100 km requires RPQ 8P2263 or 8P2340
ISC-3
ISC-M (hot-plug) ISC-D (hot-plug)
LC Duplex connector
ISC link
EP
The z114 and z196 will be the last servers to offer ordering of ISC-3. Customers should start migrating to PSIFB coupling links.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
113
Parallel Sysplex using InfiniBand (PSIFB)ready for even the most demanding data sharing workloads
� Simplify Parallel Sysplex connectivity
Do more with less
– Can share physical links by defining multiplelogical links (CHPIDs)
– Can consolidate multiple legacy links (ISC and/or ICB)– Can more easily address link constraints
• Define another CHPID to increase availablesubchannels instead of having to add physical links
� More flexible placement of systems in a data center
– 12x InfiniBand coupling links (FC 0171 HCA3-O and FC 163 HCA2-O)• Support optical cables up to 150 meters. No longer restricted to 7 meters between
System z CPCs– 1x InfiniBand coupling links (FC 0170 HCA3-O LR and FC FC 0168 HCA2-O LR)
• Use the same single mode fiber optic cables as ISC-3 and FICON/FCP for unrepeated distances of up to 10 km, and metropolitan distances with qualified DWDM solutions
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
114
HCA3 for Coupling Links� New 12x InfiniBand and 1x InfiniBand fanout cards
� Exclusive to z196 and z114
– HCA3-O fanout for 12x InfiniBand coupling links
• CHPID type – CIB
– Improved service times with 12x IFB3 protocol
– Two ports per feature
– Fiber optic cabling – 150 meters
– Supports connectivity to HCA2-O
– Link data rate of 6 GBps
– HCA3-O LR fanout for 1x InfiniBand coupling links
• CHPID type – CIB
– Four ports per feature
– Fiber optic cabling
– 10 km without repeaters, RPQ 8P2263 or 8P2340 is required for 20 km support
– Extended distances of up to 100 km are also possible using Dense Wavelength Division Multiplexors (DWDMs)
– Over 100 km requires RPQ 8P2263 or 8P2340
– Supports connectivity to HCA2-O LR
– Link data rate server-to-server 5 Gbps
– Link data rate with DWDM; 2.5 or 5 Gbps
* Performance considerations may reduce the number of CHPIDs per port
Up to 16 CHPIDs – across 2 ports*
Up to 16 CHPIDs – across 4 ports*
IFB IFB
HCA3-O for 12x IFB & 12x IFB3
IFB
HCA3-O LR for 1x IFB
IFB IFB IFB
Note: The InfiniBand link data rates of 6 GBps, 3 GBps, 2.5 Gbps, or 5 Gbps do not represent the performance of the link. The actual performance is dependent upon many factors including latency through the adapters, cable lengths, and the type of workload.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
115
HCA3-O fanout for 12x InfiniBand coupling links
• CHPID type – CIB
– Two ports per feature
– Fiber optic cabling – 150 meters
– Supports connectivity to HCA2-O
– Link data rate of 6 GBps
Two protocols (IFB & IFB3)
1. 12x IFB = HCA3-O to HCA2-O
2. 12x IFB3 = HCA3-O to HCA3-O (see below)
• Improved service times, 12x IFB3 service times are
designed to be 40% faster than 12x IFB
12x IFB3 protocol activation requirements
– Maximum of four CHPIDs per HCA3-O port
• If more than four CHIPDs are defined per port, links will run at normal 12x IFB service times
• IFB3 protocol activated as long as 4 CHPIDs or less are defined. No configuration settings required.
HCA3-O HCA3-O
Note: The InfiniBand link data rates of 6 GBps, 3 GBps, 2.5 Gbps, or 5 Gbps do not represent the performance of the link. The actual performance is dependent upon many factors including latency through the adapters, cable lengths, and the type of workload.
New 12x InfiniBand fanout cards, exclusive to z196 and z114
* Performance considerations may reduce the number of CHPIDs per port
*Up to 16 CHPIDs – across 2 ports
System z10 z196
HCA3-OHCA2-O
IFB IFB3
IMPORTANT !
Attachment to System z9 HCA1 not supported
HCA3-O 12x Details
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
116
Note: The InfiniBand link data rates of 6 GBps, 3 GBps, 2.5 Gbps, or 5 Gbps do not represent the performance of the link. The actual performance is dependent upon many factors including latency through the adapters, cable lengths, and the type of workload.
New 1x InfiniBand fanout cards, exclusive to z196 and z114
* Performance considerations may reduce the number of CHPIDs per port
z196
HCA3-O LR
System z10
HCA2-O
LR
HCA3-O LR
*Up to 16 CHPIDs – across 4 ports
HCA3-O LR fanout for 1x InfiniBand coupling links
• CHPID type – CIB
– Four ports per feature
– Fiber optic cabling
– Link sharing between multiple Sysplexes
– Dstance of up to 10 km without repeaters, RPQ 8P2263 or 8P2340 is required for 20 km support
– Extended distances of up to 100 km are also possible using Dense Wavelength Division Multiplexors (DWDMs), going over 100 km requires RPQ 8P2263 or 8P2340
– Supports connectivity to HCA2-O LR
– Link data rate server-to-server 5 Gbps
– Link data rate with DWDM; 2.5 or 5 Gbps
CHPID / Subchannel Options
– 32 subchannels per CHPID (default)
– Option to define 32 or 7 subchannels
– z114 or z196 GA2 to z114 or z196 GA2
HCA3-O LR
HCA3-O LR
HCA3-O
LR
DWDM2.5 Gbps
z196
HCA3-O LR
DWDM5 Gpbs
Server5 Gpbs
System z9(not supported)
z114TLLB116
HCA3-O LR 1x Details
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
117
InfiniBand coupling links on z196 and z114 Summary
DDR = double date rate, SDR = single data rate* For z114 carry forward only** Performance considerations may reduce the number of CHPIDsper port
Type Speed Distance Fanout Cabling
12x InfiniBand 6 or 3 GBps 150 metersHCA2-O
HCA3-O50µ MM (OM3) fiber
1x InfiniBand 5 or 2.5 Gbps 10 kmHCA2-O LR*
HCA3-O LR9µ SM fiber
� Ports exit from the front of a book/processor drawer with HCAs. Unlike ISC-3 links, does not use I/O card slots
� 12x InfiniBand – z196, z114, z10, z9– DDR at 6 GBps
• z196/z114 and z10 – SDR at 3 GBps
• z196/z114 & z10 to z9• z9 to z9 connection not supported
� 1x InfiniBand – z196/z114 and z10 (not available for z9)
– DDR at 5 Gbps (Server to Server and with DWDM)
– SDR at 2.5 Gbps (with DWDM)
HCA2-O IFB IFB
Up to 16 CHPIDs – across 2 ports
HCA2-O LR* IFB IFB
Up to 16 CHPIDs – across 2 ports
Up to 16 CHPIDs – across 2 ports**
HCA3-O IFB & IFB3 IFB & IFB3
Up to 16 CHPIDs – across 4 ports**
HCA3-O LR IFB IFB IFB IFB
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
118
InfiniBand Connectivity Options
zEnterprise to
zEnterprise orz10 (HCA2-O LR)
zEnterprise to
zEnterprise (HCA3) or
z10 (HCA2)
zEnterprise to
zEnterprise (HCA3) or
z10 (HCA2)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
119
12x InfiniBand coupling links - Multiple channel paths
� Up to 16 CHPIDs across two ports
– More subchannels per physical link – CF Receiver CHPIDs can share links– NOT more subchannels per CHPID– MIF uses same address, 7 subchannels per CHPID
CHPID FF7 subchannels
CHPID FE7 subchannels
Single 12x IFB link
14 subchannels
7 SubChannels per channel CHPID
Up to 16 channel paths across two ports
___________________
112 subchannels across two ports
Eg.)
HCA3-O
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
120
� Up to 16 CHPIDs using same physical links – More subchannels per physical link– Link sharing by different Sysplexes
� Now more subchannels per CHPID– 32 subchannels per CHPID
– Option to define 32* or 7 subchannels– z114 or z196 GA2 to z114 or z196 GA2
CHPID FF32 subchannels
CHPID FE32 subchannels
One 1x IFB link
64 subchannels
32* subchannels per CHPID (default)
Up to 16 CHPIDs per HCA3-O LR___________________
512 subchannels per HCA3-O LR
z114 and z196 GA2 1x InfiniBand Coupling LinksMultiple CHPIDs per link, 32 or 7 subchannels per CHPID (HCA2-O LR and HCA3-O LR)
HC
A3-O
LR
For Example:
Carry forward only for z114
*HCD defaults to 32 subchannels.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
121
12x InfiniBand SDR
12x InfiniBand
HCA2-O*
HCA2-C I/O Drawer
or I/0 Cage
ISC-3
ISC-3
ISC-3
ISC-3
z196, z114, z10 EC, z10 BC
IFB-MP
z196/z114
ISC-3
z196, z114, z10 EC, z10 BC, z9 EC, z9 BC
HCA2-O LR*
Up to 150 meters
Up to 150 meters
Up to 10/100 km
.... ....
HCA2-O**
.... ....
1x InfiniBand DDR
Up to 10/100KM
z9 EC and z9 BC
Fanouts
HCA3-O HCA3-O LR
.... ....
12x InfiniBand
Up to 150 meters
1x InfiniBand DDR
Up to 10/100KM
** Not supported on the z114 ** Carry forward only
z196/z114 Coupling Links
� Fanout, not I/O slot, used for InfiniBand
� ICB-4 – No longer supported
� ETR – No longer supported
� All coupling links support STP
� Sysplex Coexistence – z10 EC and BC and z9 EC and BC only
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
122
System z InfiniBand HCA Adapter Options – Summary
� Host Channel Adapter Optical Fanouts for PSIFB – z196 supports HCA3O, HCA3-O LR, HCA2-O and HCA2-O LR fanouts– z114 supports HCA3-O, HCA3-O LR and HCA2-O (carried forward only) fanouts
� 12X InfiniBand-SDR (3 GBps)– z196/z114 to z9 – Uses 12 lanes for a total link rate of 3 GBps and is a point-to-point connection, maximum length of 150 meters– New 50 micron OM3 (2000 MHz-km) multimode fiber with MPO connectors
� 12X InfiniBand-DDR (6 GBps)– z196/z114 to z196/z114– z196/z114 to z10– Uses 12 lanes for a total link rate of 6 GBps and is a point-to-point connection, maximum length of 150 meters– 50 micron OM3 (2000 MHz-km) multimode fiber with MPO connectors
� 1X InfiniBand-DDR LR (5 Gbps)– z196/z114 to z196/z114– z196/z114 to z10 – Uses one lane for a total link rate of 5 Gbps and supports an unrepeated distance of 10 km– Repeated distances at up to 100 km when attached to a Dense Wavelength Division Multiplexer (DWDM) qualified by
System z– Speed may be auto-negotiated if the attached DWDM is capable of operating at SDR or DDR
• Supports SDR at 2.5 Gbps when connected to a DWDM capable of SDR speed • Supports DDR at 5 Gbps when connected to a DWDM capable of DDR speed
– 9 micron single mode fiber optic cables with LC Duplex connectors (same cable as ISC3)
Note: The InfiniBand link data rates of 6 GBps, 3 GBps, 2.5 Gbps, or 5 Gbps do not represent the performance of the link. The actual performance is dependent upon many factors including latency through the adapters, cable lengths, and the type of workload. With InfiniBand coupling links, while the link data rate may be higher than that of ICB (12x InfiniBand (SDR or DDR) or ISC-3 (1x InfiniBand (SDR or DDR), the service times of coupling operations are greater, and the actual throughput may be less than with ICB links or ISC-3 links.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
123
Coupling links using InfiniBand Trade Association standard
� 1x InfiniBand
� LC Duplex connector
� One pair of fiber (1x) – one fiber for transmit and one fiber for receive
• 1x InfiniBand operating at 2.5 Gbps or 5 Gbps is exclusive to System z10 and z196/z114 servers
• All attachments to an outside cable plant (including public "dark fiber") are supported only through a patch panel or Wavelength Division Multiplexer (WDM) product.
• 10 to 100 km repeated with an InfiniBand DDR or SDR qualified DWDM. (STP qualification is also required if STP is in use.)
2.5 Gbps (SDR) 5 Gbps (DDR)
FiberCore (µ )
(Light source)@ wavelength
Unrepeateddistance
Opticalpassive
loss
Unrepeateddistance
Opticalpassive
loss
9µ SMLR laser
@ 1310 nm10 km
6.2 miles5.66 dB
10 km6.2 miles
5.66 dB
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
124
1x InfiniBand9/125 micrometer single mode fiber optic cabling
� Cables available from:
– IBM Global Technology Services (GTS)– Your preferred cable provider
� Fiber core – 9µ single mode
� Light source – LX laser @ wavelength: @ 1310 nm
Note: the fiber optic cabling is the same
as used with ISC-3, FICON LX, 10 GbE LR, and GbE LX
LC Duplex connector
LC Duplex harness
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
125
Coupling links using InfiniBand Trade Association standard
� 12x Multi-fiber Push-On (MPO) connector
� 24-fiber cable with Duplex MPO connectors– 12 fibers for transmit and 12 fibers for receive
1.12x InfiniBand SDR links operating at 3 GBps (2.5 Gbps per lane) are used to connect a z196/z114 to a z9 or z196/z114 to z10
2.12x InfiniBand DDR links operating at 6 GBps (5.0 Gbps per lane) are used to connect z196/z114 servers or z196/z114 to z10
3 GBps (SDR) 6 GBps (DDR)
FiberCore (µ )
(Light source)
FiberBandwidth
@ wavelength
Unrepeateddistance
Opticalpassive
loss
Unrepeateddistance
Opticalpassive
loss
50µ MM(SX laser)
2000 MHz-km@ 850 nm
150 meters492 feet
2.06 dB150 meters
492 feet2.06 dB
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
126
� Cables available from:– IBM Global Technology Services (GTS)– Anixter www.anixter.com/– Computer Crafts Inc. www.computer-crafts.com/– Tyco www.tycoelectronics.com/– Fujikura www.fujikura.com/
� Fiber core – 50u multimode
� Light source – SX laser
� Fiber bandwidth @ wavelength: 2000 MHz-km @ 850 nm
� IBM cable part numbers highly recommended
Supported 12x InfiniBand DDR cable lengthsOM3 50/125 micrometer multimode fiber optic cabling
Cable Cable
Item Cable Length Length Connector
Description IBM P/N Meters Feet Type
Duplex 24-fiber cable Assembly 41V2466 10.0 m 32.8 f MPO - MPO
Duplex 24-fiber cable Assembly 15R8844 13.0 m 42.7 f MPO - MPO
Duplex 24-fiber cable Assembly 15R8845 15.0 m 49.2 f MPO - MPO
Duplex 24-fiber cable Assembly 41V2467 20.0 m 65.6 f MPO - MPO
Duplex 24-fiber cable Assembly 41V2468 40.0 m 131.2 f MPO - MPO
Duplex 24-fiber cable Assembly 41V2469 80.0 m 262.4 f MPO - MPO
Duplex 24-fiber cable Assembly 41V2470 120.0 m 393.7 f MPO - MPO
Duplex 24-fiber cable Assembly 41V2471 150.0 m 492.1 f MPO - MPO
Duplex 24-fiber cable Assembly 42V2083 Custom N/A MPO - MPO
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
127
Non-PSIFB Coupling Links
Description F/C Ports Available Comments
ISC-D 0217 N/A New/Carry Forward Mother Card
ISC-D 0218 1 to 2 New/Carry Forward ISC-D (Daughter Card)
ISC-3 Link 0219 1 to 4 New/Carry Forward Port(s) Enabled
RPQ 8P2197 2 New/Carry Forward ISC-3 20KM
ICB-3 0993 Not Available
ICB-4 3393 Not Available
The z114 is the last server to offer ordering of new ISC-3 features.
ISC-3 requires an I/O Drawer.
Server Time Protocol (STP) Required
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
128
� Theoretical maximum rates shown� 9672 / z900 / z800 / z990 / z890 Connectivity to z196/z114 Not Supported� z196 and z114 does not support ICB connectivity� 1x InfiniBand supports single data rate (SDR) at 2.5 Gbps when connected to a DWDM capable of SDR
speed and double data rate (DDR) at 5 Gbps when connected to a DWDM capable of DDR speed or when point-to-point with another z196/z114 or z10
� System z9 does NOT support 1x InfiniBand DDR or SDR InfiniBand Coupling Links
System z – CF Link Connectivity (Peer Mode only)
Connectivity Optionsz196/z114
ISC-3
z196/z114
1x InfiniBand
z196/z114
12x InfiniBand
z196 / z114 / z10 / z9
ISC-3 (RPQ 8P2197 – 20 km)1 Gbps N/A N/A
z196 / z114 / z10 / z9
ISC-3 (10/100 km)2 Gbps N/A N/A
z9 EC or z9 BC with 12x InfiniBand SDR (150 m) N/A N/A 3 GBps
z196 / z114/ z10 1x-InfiniBand DDR(10/100 km) N/A 5 Gbps N/A
z196 / z114 /z10 12x InfiniBand DDR
(150 m)N/A N/A 6 GBps
Note: The InfiniBand link data rates of 6 GBps, 3 GBps, 2.5 Gbps, or 5 Gbps do not represent the performance of the link. The actual performance is dependent upon many factors including latency through the adapters, cable lengths, and the type of workload. With InfiniBand coupling links, while the link data rate may be higher than that of ICB (12x InfiniBand (SDR or DDR) or ISC-3 (1x InfiniBand (SDR or DDR), the service times of coupling operations are greater, and the actual throughput may be less than with ICB links or ISC-3 links.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
129
System z – Maximum Coupling Links and CHPIDs
Server1x IFB
(HCA3-O LR)
12x IFB & 12x IFB3(HCA3-O)
1x IFB(HCA2-O LR)
12x IFB (HCA2-O)
IC ICB-4 ICB-3 ISC-3Max
External Links
MaxCoupling CHPIDs
z19648
M15 – 32*
32
M15 – 16*
32
M15 – 16*
CF only
32
M15 – 16*32 N/A N/A 48 104 (1) 128
z114M10 – 32*
M05 – 16*
M10 – 16*
M05 – 8*
M10 – 12
M05 – 8*
CF only
M10 – 16*
M05 – 8*32 N/A N/A 48
M10 (2)
M05 (3) 128
z10 EC N/A N/A32
E12 – 16*
32
E12 – 16*32
16
(32/RPQ)N/A 48 64 64
z10 BC N/A N/A 12 12 32 12 N/A 48 64 64
z9 EC N/A N/A N/AHCA1-O
16S08 - 12
32 16 16 48 64 64
z9 BC N/A N/A N/AHCA1-O
1232 16 16 48 64 64
* Uses all available fanout slots. Allows no other I/O or coupling.
1. A z196 M49, M66 or M80 supports a maximum 104 extended distance links (48 1x IFB and 48 ISC-3) plus 8 12x IFB links. A z196 M32 supports a maximum 96 extended distance links (48 1x IFB and 48 ISC-3) plus 4 12x IFB links*. A z196 M15 supports a maximum 72 extended distance links (24 1x IFB and 48 ISC-3) with no 12x IFB links*.
2. z114 M10 supports a maximum of 72 extended distance links (24 1x IFB and 48 ISC-3) with no 12x IFB links*.3. z114 M05 supports a maximum of 56 extended distance links (8 1x IFB and 48 ISC-3) with no 12x IFB links*.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
130
IBM zEnterprise 114 (z114)
I/O Connectivity Summary
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
131
z114 I/O Connectivity
FeaturesOffered
AsMaximum # of features
Maximumchannels
Incrementsper feature
Purchaseincrements
ESCON NB 16 240 channels 1 - 15 active 4 channels
FICON
FICON Express8S NB 64 128 channels 2 channels 2 channels
FICON Express8 CF* 16 64 channels 4 channels 4 channels
FICON Express4 CF 16 64 channels 4 channels 4 channels
FICON Express4-2C CF* 16 32 channels 2 channels 2 channels
ISC-3 NB 12 48 links 4 links 1 link
OSA-Express
OSA-Express4S 10 GbE, GbE NB 48 96 ports (48 for 10 GbE) 1 (10 GbE) / 2 ports 1 feature
OSA-Express3 1000BASE-T NB 16 64 ports 4 ports 1 feature
OSA-Express3-2P 1000BASE-T NB 16 32 ports 2 ports 1 feature
OSA-Express3 10 GbE, GbE CF* 16 64 ports (32 for 10 GbE) 2 (10 GbE) / 4 ports 1 feature
OSA-Express3-2P GbE CF* 16 32 ports 2 ports 1 feature
OSA-Express2** CF 16 32 ports 2 ports 1 feature
Crypto
Crypto Express3 NB 8 16 PCIe adapters 2 PCIe adapters 2 features***
Crypto Express3-1P NB 8 8 PCIE adapters 1 PCIe adapter 2 features***
* Can be carried forward or ordered on MES using RPQ 8P2534** OSA-Express2 10 GbE LR is not supported as a carry forward*** Two features initially, one thereafter
NB = New BuildCF = Carry Forward
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
132
z114 Coupling Link connectivity summary
FeaturesMinimum # of features
Maximum # of features
Maximum connections
Incrementsper feature
Purchase increments
ISC-3 0 12 48 links 4 links 1 link
HCA2-O LR* (1x) 04 – M05**6 – M10**
8 links – M0512 links – M10
2 links 2 links
HCA2-O (12x) 04 – M05**8 – M10**
8 links – M0516 links – M10
2 links 2 links
HCA3-O LR (1x) 04 – M05**8 – M10**
16 links – M0532 links – M10
4 links 4 links
HCA3-O (12x) 04 – M05**8 – M10**
8 links – M0516 links – M10
2 links 2 links
* Carry forward only ** Uses ALL the available fanout slots for PSIFB. Allows no other I/O or coupling.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
133
IBM zEnterprise
Server Time Protocol Support and Enhancements
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
134
Glossary for System z Server Time Protocol (STP)
Acronym Full name Comments
Arbiter ArbiterServer assigned by the customer to provide additional means for the Backup Time Server to determine whether it should take over as the Current Time Server.
BTS Backup Time ServerServer assigned by the customer to take over as the Current Time Server (stratum 1 server) because of a planned or unplanned reconfiguration.
CST Coordinated Server TimeThe Coordinated Server Time in a CTN represents the time for the CTN. CST is determined at each server in the CTN.
CTN Coordinated Timing NetworkA network that contains a collection of servers that are time synchronized to CST.
CTN IDCoordinated Timing Network Identifier
Identifier that is used to indicate whether the server has been configured to be part of a CTN and, if so, identifies that CTN.
CTS Current Time ServerA server that is currently the clock source foran STP-only CTN.
Going Away SignalA reliable unambiguous signal to indicate that the CPC is about to enter a check stopped state.
PTS Preferred Time ServerThe server assigned by the customer to be the preferred stratum 1 server in an STP-only CTN.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
135
Current STP Recovery – in a nut shell
� CANNOT have two Stratum 1 servers in timing network
� Backup Time Server (BTS) can take over as Current Time Server (CTS), active Stratum 1, only if either:
– Preferred Time Server (PTS) can indicate it has “failed”– BTS can unambiguously determine the PTS has “failed”
� Coordinated Timing Network (CTN) with only 2 servers/CFs– Based on combination of:
• Server Offline Signal (OLS- Channel going away signal) and • Console Assisted Recovery (CAR)
– Server Offline signal (OLS) transmitted on a channel by the server to indicate that the channel is going offline
– CAR initiated when OLS not received• Uses HMC/SE LAN path to determine if BTS can take over as CTS
� Coordinated Timing Network (CTN) with 3 or more servers/CFs– If BTS loses communication on all established paths to CTS
• BTS and Arbiter communicate to establish if Arbiter also has lost communication on all established paths to CTS
– If both BTS and Arbiter cannot communicate with CTS• BTS takes over as CTS (S1)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
136
Server Time Protocol (STP) Recovery Enhancements
�The new generation of host channel adapter, HCA3-O
�Going away signal
�Server on which it is running is about to enter a failed (check stopped) state.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
137
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
138
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
139
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
140
How it works
• Going Away Signal sent by the Current Time Server (CTS) in an STP-only Coordinated Timing Network (CTN)
• Received by the Backup Time Server (BTS)
• BTS can safely take over as the CTS
• without relying on Offline Signal (OLS) in a 2 server CTN or theArbiter in a CTN with 3 or more servers
• Current recovery design is still used when going away signal is not received by BTS and for other failures
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
141
Pre-reqs
• This enhancement is only available if you have a HCA3-O on the CTS communicating with a HCA3-O on the BTS.
• The STP recovery design is still available for the cases when a Going Away Signal is not received or for other failures besides a server failure
• Infiniband (IFB) links using • HCA3-O to HCA3-O 12x IFB to 12x IFB • HCA-O LR to HCA3-O LR 1x IFB
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
142
Dependencies
• Dependencies on OLS and CAR removed in a 2 server CTN –Going Away Signal has priority over OLS in a 2 server CTN
• Dependencies on BTS-->Arbiter communication removed in CTNswith 3 or more servers
–BTS can also use going away signal to take over as CTS for CTNswith 3 or more servers without communicating with Arbiter
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
143
Improved time coordination for virtual servers in zBX
� Network Time Protocol (NTP) clients running on blades and appliances in zBX synchronize their time to the NTP server provided by the Support Element (SE) every 2 to 4 hours
� It is important for the SE’s clock to maintain good time accuracy
� this enhancement provides the capability for the components in zBXto maintain an approximate time accuracy of 100 milliseconds to an NTP server if they synchronize to the SE’s NTP server at least once an hour
� The IBM WebSphere DataPower Integration Appliance XI50 for zEnterprise (DataPower XI50z) synchronizes to the SE’s BOC once an hour.
� The POWER7 blade and Optimizer synchronize to the SE every 4 hours
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
144
SE Enhancement
� An enhancement has been made to improve the time accuracy of theSE’s Battery Operated Clock (BOC) by synchronizing the SE’s BOC to the server’s Time of Day Clock (TOD) every hour, instead of the previous synchronization which took place every 24 hours.
� This enhancement will allow the SE’s clock to maintain a time accuracy of 100 milliseconds to an NTP server configured as the External Time Source in an STP-only CTN.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
145
SE BOC Update
� At IML SE BOC used to set the CPC TOD
� Prior DR93, Primary SE synchronizes with CPC TOD every 24 hours since SE BOC may have a significantly larger accumulated drift in time compared to CPC TOD
� Could result in repeated timestamps in logs
� To ensure that new Primary SE running with accurate CEC time–Primary SE BOC synchs with the CPR TOD –Alternate SE BOC synchs with Primary SE BOC –Alternate SE switches to Primary SE
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
146
zBX Virtual Servers improved time
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
147
NTP thresholds
� Allows you to specify threshold settings in order to suppress generation of hardware and operating system messages related to changes in the NTP server stratum level or source ID.
� Select the stratum level threshold value to indicate the NTP server stratum level that must be reached before generating messages.
� Select the source ID time threshold value to indicate the amount of time that must pass before a change in the source ID generates messages.
– Messages issued if the source ID does not return to the original value within the specified time period.
� Example:– If NTP server stratum level changes
from 2 to 3 and back from 3 to 2, no messages will be generated
– If NTP server stratum level changes from 2 to 4, messages will be generated
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
148
STP Server Role Assignments (3 or more servers)
• STP server role assignments should be accounted for and included in ALL installation / migration planning activities !
• Process documented in techdoc WP101833 should be followed any time a server is take down for planned or unplanned outages. Failure to follow these procedures can result in a sysplex wide outage!
• Server role reassignments are easy and can be changed concurrently. Roles assignments performed under the “Network Configuration”panel via the “System (Sysplex) Time” task.
• All server roles assignments must be done from the CPC that is assigned as the PTS.
– For PTS reassignments only, the PTS assignment to another CPC must be done from the CPC that will become the new PTS.
New whitepaper released - WP101833http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP101833
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
149
References for Additional Information
� Redbooks– Server Time Protocol Planning Guide, SG24-7280– http://www.redbooks.ibm.com/redpieces/abstracts/sg247280.html– Server Time Protocol Implementation Guide, SG24-7281– http://www.redbooks.ibm.com/redpieces/abstracts/sg247281.html
� Education– Introduction to Server Time Protocol (STP)
• Available on Resource Link at General Availability (GA)• www.ibm.com/servers/resourcelink/hom03010.nsf?OpenDatabase
� STP Web site– www.ibm.com/systems/z/pso/stp.html
� Systems Assurance– The IBM team is required to complete a Systems Assurance Review (SAPR Guide, SA06-012) and to
complete the Systems Assurance Confirmation Form via Resource Link– http://w3.ibm.com/support/assure/assur30i.nsf/WebIndex/SA779
� For further information on NTP and the NTP Public services project, refer to the Web sites:– http://www.ntp.org– http://support.ntp.org– http://www.faqs.org/rfcs/rfc1305.html
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
150
IBM zEnterprise
Cryptography
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
151
� Cryptographic Coprocessor Facility – Supports “Secure key” cryptographic processing � PCICC Feature – Supports “Secure key” cryptographic processing � PCICA Feature – Supports “Clear key” SSL acceleration� PCIXCC Feature – Supports “Secure key” cryptographic processing � CP Assist for Cryptographic Function allows limited “Clear key” crypto functions from any CP/IFL
– NOT equivalent to CCF on older machines in function or Crypto Express2 capability� Crypto Express2 – Combines function and performance of PCICA and PCICC� Crypto Express3 – PCI-e Interface, additional processing capacity with improved RAS
2001 2002 2003 2004 20062005 2007 2008 2010/112009
Crypto Express3 z10 EC/BC z196/z114
z9 EC z9 BC z10 EC/BCCrypto Express2
Cryptographic Coprocessor Facility (CCF)
PCI Cryptographic Coprocessor (PCICC)
PCI Cryptographic Accelerator (PCICA)
PCIX Cryptographic Coprocessor (PCIXCC)
CP Assist for Cryptographic Functions
System z Crypto History
z990/z890
G3, G4, G5, G6, z900, z800
G5, G6, z900, z800
z800/z900
z9 EC z9 BC z10 EC/BCz990 z890
z990
z990
z890
z890
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
152
z196/z114 Cryptographic Enhancements
� Elliptic Curve Cryptography Digital Signature Algorithm (ECC), an emerging public key algorithm expected eventually to replace RSA cryptography in many applications. ECC is capable of providing digital signature functions and key agreement functions. The new CCA functions provide ECC key generation and key management and provide digital signature generation and verification functions compliance with the ECDSA method described in ANSI X9.62 "Public Key Cryptography for the Financial Services Industry: The Elliptic Curve Digital Signature Algorithm (ECDSA)". ECC uses keys that are shorter than RSA keys for equivalent strength-per-key-bit; RSA is impractical at key lengths with strength-per-key-bit equivalent to AES-192 and AES-256. So the strength-per-key-bit is substantially greater in an algorithm that uses elliptic curves.
� ANSI X9.8 PIN security which facilitates compliance with the processing requirements defined in the new version of the ANSI X9.8 and ISO 9564 PIN Security Standards and provides added security for transactions that require Personal Identification Numbers (PIN).
� Enhanced Common Cryptographic Architecture (CCA), a Common Cryptographic Architecture (CCA) key token wrapping method using Cipher Block Chaining (CBC) mode in combination with other techniques to satisfy the key bundle compliance requirements in standards including ANSI X9.24-1 and the recently published Payment Card Industry Hardware Security Module (PCI HSM)standard.
� Secure Keyed-Hash Message Authentication Code (HMAC), a method for computing a message authentication code using a secret key and a secure hash function. It is defined in the standard FIPS 198, "The Keyed-Hash Message Authentication Code". The new CCA functions support HMAC using SHA-1, SHA-224, SHA-256, SHA-384, and SHA-512 hash algorithms. The HMAC keys are variable-length and are securely encrypted so that their values are protected.
� Modulus Exponent (ME) and Chinese Remainder Theorem (CRT), RSA encryption and decryption with key lengths greater than 2048-bits and up to 4096-bits.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
153
Summary of Cryptographic Enhancements (2011)
� Crypto Express3 and Crypto Express3 -1P (z114 only)
� AES Key Encrypting Keys (AES KEKs) and AES typed keys, in general
� TR-31 key wrapping method for interoperable secure key exchange
� Elliptic Curve Cryptography (EC-DH key agreement protocol)
� PIN Decimalization table protection
� (Rivest-Shamir-Adleman) RSA-Optimal Asymmetric Encryption Padding (OAEP) with SHA-256
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
154
TKE 7.1 LIC
� Support of new hardware, firmware, software
� Support requirements for standards – Includes using stronger/newer types of encryption inside the TKE
� Simplification of tasks
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
155
TKE 7.1 LIC features
Security
� New smart card reader support
� Ease of Use/Simplification
� Key Data Migration Wizard update – A single Process for Loading an Entire Key
Hardware Support
� New Operational AES Key type – Users can load and manage the new AES EXPORTER,
IMPORTER, and CIPHER operational keys from the TKE’s crypto module notebook
� PIN Decimalization Table Support– CCA version 4.2 for the Crypto Express3 feature
includes support for 100 decimalization tables for each domain on a host cryptographic adapter.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
156
TKE 7.1 LIC features
RAS
� HW Crypto module Status Display – Users will be able to display the current status of the HSM that is being managed – Users in privileged access mode will be able to Display Active IDs on the TKE Console
� Dynamic PKA Master Key Change Support
Standards/ Compliance
� PCI HSM Standards Compliance – design
� Configuration Profile Support
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
157
Description F/C Engines Available Comments
CPACF enablement 3863 N/A New/Carry Forward No Charge – DES/TDES
Crypto Express3 0864 2 New/Carry Forward
Crypto Express3-1P 0871 1 New z114 /Carry Forward z10 BC and z114 only
TKE 7.0 LIC 0860 N/A New/Carry Forward Upgrade available to TKE 7.1
TKE 7.1 LIC 0867 N/A New Required for z196/z114
TKE Workstation 0841 N/A New/Carry ForwardRequired for z196/z114 and ships with 7.1 LIC
Smart Card Reader 0885 N/A New/Carry Forward (7.0)EC to change from NXP JCOP 41 to NXP JCOP 31
Smart Card 0887 N/A Yes
Additional S-Cards 0884 N/A New/Carry Forward10-pack. Requires TKE 7.0 LIC or higher.
Summary of Crypto Features for z196/z114
Note: Crypto Express3 requires a I/O Cage (z196 only) or I/O Drawer (FC4000)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
158
System z Evaluations & Certifications
The Common Criteria
program establishes
an organizational and
technical framework
to evaluate the
trustworthiness of IT
Products and
protection profiles
z/OS z/VM
Linux Linux Linux
Virtualization with partitions
Cryptography
• Common Criteria EAL4+ • with CAPP and LSPP• z/OS 1.7 + RACF• z/OS 1.8 + RACF• z/OS 1.9 + RACF• z/OS 1.10 + RACF (OSPP)• z/OS 1.11 + RACF (OSPP)
• z/OS 1.10 IPv6 Certification by JITC
• IdenTrust™ certification for z/OS PKI Services
• FIPS 140-2• z/OS System SSL 1.10
z/OS System SSL 1.11• z/OS ICSF PKCS#11
Services – z/OS 1.11• Statement of Integrity
z/OS
• System z9 EC and z9 BC System z10 EC and z10 BC
• Common Criteria EAL5 with specific target of evaluation -- LPAR: Logical partitions
• System z196• Common Criteria EAL5+ with specific target of
Evaluation -- LPAR • z114 pending
• Crypto Express2 & Crypto Express3 Coprocessors- FIPS 140-2 level 4 Hardware Evaluation- Approved by German ZKA
• CP Assist- FIPS 197 (AES) - FIPS 46-3 (TDES)- FIPS 180-3 (Secure Hash)
� Common Criteria
� z/VM 5.3
• EAL 4+ for CAPP/LSPP
• System Integrity Statement
• z/VM 6.1 Under evaluation
z/VM
� Common Criteria
� SUSE SLES10 certified at EAL4+ with CAPP
� Red Hat EL5 EAL4+ with CAPP and LSPP
� OpenSSL - FIPS 140-2 Level 1 Validated
� CP Assist - SHA-1 validated for FIPS 180-1 -DES & TDES validated for FIPS 46-3
Linux on System z
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
159
IBM zEnterprise 114 (z114)
RAS
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
160
Scheduled (CIE+Disrupt ive Pat ches + ECs)
Planned - (MES + Dr iver Upgrades)
Unscheduled (UIRA)
Sources of Outages - Pre z9 -Hrs/Year/Syst- Prior
ServersIBM System z9® IBM System z10™ zEnterprise
Unscheduled Outages
Scheduled Outages
Planned Outages
Preplanning requirements
Power & Thermal
Management
Increased Focus over time
� � �
� ��
� �
�
�
�
�
�
Temperature = Silicon Reliability Worst EnemyWearout = Mechanical Components Reliability Worst Enemy.
IBM System z Overall RAS Strategy…..Continuing our RAS focus helps avoid outages
Imp
ac
t of O
uta
ge
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
161
Preventing All Outages
� Unscheduled Outages
– Advanced Memory RAIM design
• Enhanced Reed-Solomon code (ECC) – 90B/64B
• Protection against Channel/DIMM failures
• Chip marking for fast DRAM replacements
– Mirrored Key cache
– ECC on all Store-in caches (including new EDRAM caches)
– Improved SER resistant latches
– FCP end-to-end checking at OS level
– Integrated TCP/IP checksum generation/checking
– Enhanced STP take over architecture
– Integrated EPO switch cover (protecting the switch during repair actions)
– Continued focus on Firmware
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
162
Preventing All Outages
� Scheduled Outages
– Double memory data bus lane sparing (reducing repair actions)
– Single memory clock bus sparing
– Double DRAM chipkill tolerance
– Field Repair of interface between processor chip and cache chip and between cache chips (fabric bus)
– Fast bitline delete on L3/L4 cache (largest caches)
– Redundant (N+2) Humidity Sensors
– Redundant (N+2) Altimeter Sensors
– PCIe based I/O infrastructure with Failover capability
– Concurrent replace of existing and new PCIe I/O drawer
– Concurrent blower repair on PCIe I/O drawer
– Light strip to guide repairs on PCIe IO drawer
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
163
Preventing All Outages
� Planned Outage / Preplanning
– Concurrent add of existing and new PCIe I/O drawer
– Eliminate disruption on Cryptography Adapter (level 3 and higher) patches
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
164
Improving Reliability
� Power and Thermal Optimization and Management
– Static Power Save Mode under z/OS control
– FICON dynamic sub-chip level mini power down
– Smart blower management by sensing altitude and humidity
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
165
z114 – Memory Overview
Ch4Ch3
Ch2
Ch1
AS
IC
Ch0
DIMM
CLKDiff
CLKDiff
CRC
CRC
DRAM
X
X
X
X
MCU1
EC
C
RAIM
XX
X
Layers of Memory Recovery� ECC
– Powerful 90B / 64B Reed Soloman code
� DRAM Failure
– Marking technology; no half sparing needed
– 2 DRAM can be marked
– Call for replacement on third DRAM
� Lane Failure
– CRC with Retry
– Data – lane sparing
– CLK – RAIM with lane sparing
� DIMM Failure (discrete components, VTT Reg.)
– CRC with Retry
– Data – lane sparing
– CLK – RAIM with lane sparing
� DIMM Controller ASIC Failure
– RAIM Recovery
� Channel Failure
– RAIM Recovery
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
166
DATA
CHECK
DATA
CHECKECC
Cache
Key Cache
MCU 1
Key Cache
MCU 2
16B 16B
IBM System z10 BC Memory Structure
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
167
DATA
CHECK
DATA
CHECKECCRAIM Parity
Cache
Key Cache
MCU 0
Key Cache
MCU 1
Extra column provides RAIM
function
z114 Industry Exclusive RAIM Memory Structure
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
168
� 2 drawer configuration
� New Flex cable interconnect
� Clocked from primary drawer redundant oscillators via cable and pass-thru card
� Concurrent drawer add & repair not supported
� Limited degrade capabilities
� 2 dedicated IBM spares
� PU assignment and sparing across drawers
PUPUPU PU PU PU PU
I/O Hub I/O Hub I/O Hub I/O Hub
PUPUPU PU PU PU PU
I/O Hub I/O Hub I/O Hub I/O Hub
Primary Drawer
Secondary Drawer
osc
z114 Model M10
osc
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
169
processor
Failover
I/O Mux
I/O Hub I/O Hub
I/O Mux I/O Mux I/O Mux
Failover
I/O
a
dap
ter.
I/O
a
dap
ter.
Ne
two
rk
ad
ap
ter.
Ne
two
rk a
dap
ter.
Cry
pro
Acce
lera
tor
Cry
pto
Acce
lera
tor
ISC
L
inks
ISC
Lin
ks
Alternate Path
I/O Switch
ControlUnit
I/O Switch
ControlUnit
NetworkSwitch
End User
NetworkSwitch
EndUser
z114 RAS Design ……..Fully Redundant I/O Subsystem – of existing IO cage and drawers
� Fully Redundant I/O Design– SAP / CP sparing– SAP Reassignment– I/O Reset and Failover– I/O Mux Reset / Failover– Redundant I/O Adapter– Redundant Network Adapters– Redundant Crypto Accelerators– I/O Switched Fabric– Network Switched/Router Fabric– High Availability Plugging Rules– Channel Initiated Retry– High Data Integrity Infrastructure– Type 2 Recovery– I/O Alternate Path– Network Alternate Path
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
170
processor
Failover
I/O Mux
I/O Hub I/O Hub
I/O Mux I/O Mux I/O Mux
Failover
I/O
a
dap
ter.
I/O
a
dap
ter.
Ne
two
rk
ad
ap
ter.
Ne
two
rk a
dap
ter.
Cry
pro
Acce
lera
tor
Cry
pto
Acce
lera
tor
ISC
L
inks
ISC
Lin
ks
Alternate Path
I/O Switch
ControlUnit
I/O Switch
ControlUnit
NetworkSwitch
End User
NetworkSwitch
EndUser
z114 RAS Design ……..Fully Redundant I/O Subsystem – of new PCIe drawer
� Fully Redundant I/O Design– SAP / CP sparing– SAP Reassignment– I/O Reset and Failover– I/O Mux Reset / Failover– Redundant I/O Adapter– Redundant Network Adapters– Redundant Crypto Accelerators– I/O Switched Fabric– Network Switched/Router Fabric– High Availability Plugging Rules– Channel Initiated Retry– High Data Integrity Infrastructure– Type 2 Recovery– I/O Alternate Path– Network Alternate Path
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
171
Front
Back
Domain
0
Domain
2
Domain
3
Domain
1
Failover is
Front
To
Back
Hub cards on processor
drawer
FS
P-1
ca
rd
AMD 1 AMD 2
. . . . . . . . . . . . . . . . . . .
Front View – Conceptual packaging
Classic or Native I/O Adapters
PCIe Switch Cards
Light Strip
z114 PCIe IO Drawer
� Drawer Technology
– Concurrent Adapter Repair
– Light strip for adapter service and upgrades
– Concurrent Switch card Repair
– Concurrent DCA repair
– Concurrent FSP repair
– Concurrent Drawer Add
� New Technology
– Concurrent Drawer Repair
– Concurrent upgrade from legacy drawer to new PCIedrawer
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
172
zEnterprise HMC IntegrationSystem z brings Mainframe Systems Management and Service to the Blade Environment
HMC
Ethernet
� Familiar IBM BladeCenter attached through Ethernet on z196/z114
� System z HMC style operation and control of blades
� Redundant HMC with mirroring of critical ensemble data and switchover
� Dynamic provisioning of blade resources
� Trained IBM Service personnel for service and installation. Directed repair via Primary HMC
� Call home on error
� Dynamic failure trend monitoring
� Firmware management and Upgrade
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
173
z114 RAS
� The z114 Server continues to reduce customer down time by attack all sources of
outages: unscheduled outages, scheduled outages and planned outages. Power and cooling requirements were reduced while still managing reliability.
– Major new Memory design for preventing outages
– Introducing new IO drawers, IO technologies with full concurrent service
– Introducing System z management to the mixed computing environment
– Delivering Green functions and RAS together
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
174
IBM zEnterprise 114 (z114)
Capacity on Demand
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
175
CoD Offerings
� On-line Permanent Upgrade– Permanent upgrade performed by customer (previously referred to Customer Initiated Upgrade - CIU)
� Capacity Backup (CBU) – For disaster recovery– Concurrently add CPs, IFLs, ICFs, zAAPs, zIIPs, SAPs– Pre-paid
� Capacity for Planned Event (CPE)– To replace capacity for short term lost within the enterprise due to a planned event such as a facility upgrade or
system relocation– Predefined capacity for a fixed period of time (3 days)– Pre-paid
� On/Off Capacity on Demand (On/Off CoD) – Production Capacity to satisfy periods of peak demand for computing resources– Supported through software offering – Capacity Provisioning Manager (CPM)– Payment:
• Post-paid or Pre-paid by purchase of capacity tokens• Post-paid with unlimited capacity usage• On/Off CoD records and capacity tokens configured on Resource Link
� Customer Initiated Upgrade (CIU)– Process/tool for ordering temporary and permanent upgrades via Resource Link– Permanent upgrade support:
• Un-assignment of currently active capacity• Reactivation of unassigned capacity• Purchase of all PU types physically available but not already characterized • Purchase of installed but not owned memory
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
176
Basics of CoD
On/Off CoD with tokensNo expirationCapacity- MSU %- # Engines
Tokens- MSU days- Engine days
Pre-paidCPE
On/Off CoD180 days expirationCapacity
- MSU %- # Engines
Using pre-paid unassigned
capacity up to the limit of the HWMNo expirationCapacity- MSU %
- # Engines
Post-paid
Permanent Upgrade
CBU
Capacity on Demand
Temporary Upgrade
Billable Capacity
(On/Off CoD)Replacement Capacity
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
177
Capacity on Demand Provisioning Architecture
Base Model
Dormant Capacity
Purchased Capacity
Change permanent capacity via CIU or MES order
Up to 8 records installed and active on the System and up to 200 records staged on the SE
Enforce Terms and Conditions and physical model limitations
Customer defined policy or user commands
CPM (z/OS 1.9 or higher)
APIHMC application
Query Activation
Authorization Layer
Manual operations
* Only one On/Off CoDrecord can be active
CBU – CPE – On/Off CoD
R1* R2* R3* R4* R5* R6* R7* R8*
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
178
Capacity on Demand Enhancements
z10 z196/z114
Unassigned engine purchase via CIU
Auto replenishment of On/Off CoDrecords
Manufacturing install of up to 4 CoDrecords with system ship. This is for new z196s/z114s only
On/Off CoD Administrative tests
Separate orders for purchase of unassigned engines
On/Off CoD records must be replenished manually
CoD records staged on machine deliver
No On/Off CoD administrative test
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
179
System z Capacity Resource Availability by Server
Capacity on Demand Server
Capacity BackUp z196, z114, z10 EC, z9 EC, z9 BC, z990, z890, z900, z800
Capacity Upgrade on Demand z196, z114, z10 EC, z9 EC, z9 BC, z990, z890, z900, z800
Customer Initiated Upgrade z196, z114, z10 EC, z9 EC, z9 BC, z990, z890, z900, z800
On/Off Capacity on Demand z196, z114, z10 EC, z9 EC, z9 BC, z990, z890
Capacity for Planned Event z196, z114, z10 EC, z10 BC
Capacity BackUp Capacity Upgrade on Demand
Customer Initiated Upgrade – for
ordering
On/Off Capacity on Demand
Capacity for Planned Event
z900Yes (CP only) Yes (CP, I/O, IFL, ICF,
Memory)Yes (CP, IFL, ICF, Memory)
No No
z800 Yes (CP only) Yes (CP, I/O, IFL, ICF) Yes (CP, IFL, ICF) No No
z890/z990 Yes (CP only) Yes (CP, I/O, IFL, ICF, zAAP, Memory*)
Yes (CP, IFL, ICF, zAAP, Memory*)
Yes (CP, IFL, ICF, zAAP)
No
z9Yes (CP, ICF, IFL zAAP, zIIP)
Yes (CP, I/O, IFL, ICF, zAAP, zIIP, Memory*)
Yes (CP, IFL, ICF, zAAP, zIIP, Memory*)
Yes (CP, IFL, ICF, zAAP, zIIP)
No
z10/z196/z114 Yes (CP, ICF, IFL zAAP, zIIP, SAP)
Yes (CP, I/O, IFL, ICF, zAAP, zIIP, SAP, Memory*)
Yes (CP, IFL, ICF, zAAP, zIIP, SAP, Memory*)
Yes (CP, IFL, ICF, zAAP, zIIP, SAP)
Yes (CP, IFL, ICF, zAAP, zIIP SAP)
* Not supported for some memory upgrades
Note: Upgrades are non-disruptive only where there is sufficient hardware resource available and provided pre-planning has been done
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
180
IBM zEnterprise 114 (z114)Capacity and Performance Planning
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
181
z114 Capacity Matrix
z114
� Granularity levels similar to z10 BC to facilitate upgrades and incremental growth
� Nomenclature: XYY
– X = Capacity level– YY= Number of processors – A00 = ICF or IFL only
� Any to any capacity upgrade/downgrade capability within the Model
� CBU capability from smallest to largest capacities within the Model
� On/Off CoD within the Model
� Linux only and ICF only servers
� Model M10 provides specialty engine scale out capabilities
Additional engines available on the M10
Z01 Z02 Z03 Z04 Z05
Y01 Y02 Y03 Y04 Y05
X01 X02 X03 X04 X05
W01 W02 W03 W04 W05
V01 V02 V03 V04 V05
U01 U02 U03 U04 U05
T01 T02 T03 T04 T05
S01 S02 S03 S04 S05
R01 R02 R03 R04 R05
Q01 Q02 Q03 Q04 Q05
P01 P02 P03 P04 P05
O01 O02 O03 O04 O05
N01 N02 N03 N04 N05
M01 M02 M03 M04 M05
L01 L02 L03 L04 L05
K01 K02 K03 K04 K05
J01 J02 J03 J04 J05
I01 I02 I03 I04 I05
H01 H02 H03 H04 H05
G01 G02 G03 G04 G05
F01 F02 F03 F04 F05
E01 E02 E03 E04 E05
D01 D02 D03 D04 D05
C01 C02 C03 C04 C05
B01 B02 B03 B04 B05
A01 A02 A03 A04 A05
1-way 2-way 3-way 4-way 5-way
SpecialtyEngine
SpecialtyEngine
SpecialtyEngine
SpecialtyEngine
SpecialtyEngine
SpecialtyEngine
SpecialtyEngine
SpecialtyEngine
SpecialtyEngine
SpecialtyEngine
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
182
z114 Sub-capacity Processor Granularity
A B C D E F G H I J K L M N O P Q R S TU V
W XY
Z
� The z114 has 26 CP capacity levels (26 x 5 = 130)
– Up to 5 CPs at any capacity level
• All CPs must be the same capacity level
� The one for one entitlement to purchase one zAAPand/or one zIIP for each CP purchased is the same for CPs of any speed.
– All specialty engines run at full speed
– Processor Unit Value for IFL = 100
Number of z114
CPsBase Ratio Ratio z10 BC
to z114
1 CP z10 BC Z01 1.182 CPs z10 BC Z02 1.163 CPs z10 BC Z03 1.14
4 CPs z10 BC Z04 1.13
5 CPs z10 BC Z05 1.12
Capacity level # Engines
PCI – Processor Capacity Index
5-Way4-Way
3-Way2-Way
1-Way
1-way
(sub-capacity26 PCIs)
5-way 3139 PCIs
FULL size
Specialty Engine
1-way782 PCIs
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
183
Don’t use “one number” capacity comparisons!Work with IBM technical support for capacity planning!
Customers can now use zPCR
?HiperDispatch
GHz
MIPs
tables
MSUs
ratings
The IBM Processor Capacity Reference (zPCR) is a free tool available for download that can be used to size your System z processors.
http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS1381
System z Capacity Planning in a nutshell
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
184
What’s Changing for z114
� Single set of LSPR data, based on z/OS 1.11
� Multi-Image LSPR Table (typical partition configurations assumed based on N-way)– Supports up to 5 General Purpose CPs on z114– Supports up to 80 General Purpose CPs (z196)
� Single-Image LSPR Table (single partition configurations assumed)– Supports up to 5 General Purpose CPs or 10 IFLs in single image on z114– Supports up to 80 General Purpose or IFL CPs in a single image (z196) *
� LSPR workloads – Three hardware-characteristic-based workload categories– Low, Average, and High categories, based on “Relative Nest Intensity”
• Previous workload primitives and Mixes were replaced
� Reference Processor Typical Setting– 2094-701 @ 593 MIPS for a single-partition configuration– Equates to 560 MIPS for a 5-partition configuration (multi-image LSPR table)
� Capacity planning tools affected– zPCR, zCP3000, BWATOOL, zPSG, zTPM
* Note: Whenever considering a z/OS partition with more than 48 CPs in a single partition, send a note to [email protected] for guidance.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
185
LSPR Tables
� Multi-image (MI) Processor Capacity Ratio table
– Average complex LPAR configuration for each model based on customer profiles– Most representative for vast majority of customers– Same workload assumed in every partition– z/OS only– Use for “high level” sizing– Used to develop the MSU rating
� Single-image (SI) Processor Capacity Ratio table *
– One z/OS partition equal in size to N-way of model (up to 80-way)– This table is ONLY relevant for truly single image z/OS cases– Includes z/VM and Linux workloads on both GP and IFL CPs
*Note: Likely to be dropped in a future release of zPCR
� Single-Image table is used as the basis for IBM capacity planning tools such as zPCR’s LPAR Configuration Capacity Planning function.
– Multi-Image table is a generalization of relative capacity, and is not useful for capacity planning purposes.
� Capacity for z/OS on z10 and later processors is represented with HiperDispatchturned ON.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
186
LSPR Versions Supported
� LSPR Data (supporting z114)
– z114, and IBM System z (measured in LPAR-mode)• z/OS 1.11 (Low, Average, High)*• z/VM (High/LV)• Linux (WASDB/L)
– zPCR supplements published workload categories• Low-Avg• Avg-High
� Prior LSPR Data and Tables– Removed from zPCR July 2010
• (z/OS 1.9, z/OS 1.8, z/OS 1.6 and z/OS 1.4)
� All processor data prior to System z (G6 and before)
– Removed from zPCR November 2009– Removed from zCP3000 May 2010
� All capacity planning results will be based on the most recently published data
� *Note: Measured using HiperDispatch; applies to z10, z196 and z114 only
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
187
LPAR configuration Considerations
� z/OS 1.11– Supports up to 80 CPs– LSPR data for 1 through 80 CPs assumes a single partition– Supports zAAP/zIIP Logical CPs– Measured using HiperDispatch*
• Minimizes cross book effects• Attempts to keep important Logical CPs on same Real CPs• Improves high speed buffer efficiency
� Other z/OS releases may be specified– z/OS 1.4 through z/OS 1.10 and z/OS 1.12
• z/OS 1.11 LSPR capacity data will be used• Enforce the SCP rules
� z/VM 5.4– Supports up to a maximum of 32 CPs– LSPR measured up to 32 CPs
� Linux– zPCR allows up to up to a maximum of 32 CPs– LSPR measured up to 16 CPs
� z/VSE– zPCR allows up to a maximum of 4 CPs
� CFCC– Supports up to a maximum of 16 CPs
*Note: ”Capacity Planning for z10 Upgrades” document is available TD104738
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
188
New LSPR Workload Categories
� Various combinations of prior workload primitives are measured on which the new
workload categories are based– Applications include CICS, DB2, IMS, OSAM, VSAM, WebSphere, COBOL, utilities
� Low (relative nest intensity)– Workload curve representing light use of the memory hierarchy– Similar to past high scaling workload primitives
� Average (relative nest intensity)– Workload curve expected to represent the majority of customer workloads– Similar to the past LoIO-mix curve
� High (relative nest intensity)– Workload curve representing heavy use of the memory hierarchy– Similar to the past DI-mix curve
� zPCR extends published categories– Low-Avg
• 50% Low and 50% Average– Avg-High
• 50% Average and 50% High
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
189
Choosing a Workload Category
� Average is the default
– when no other information is available
� Based on DASD I/O Rate
– From RMF Workload Activity report – Captured in EDF using CP3KEXTR* (one file per partition)
� Based on SMF 113 Data (alternate view currently provided as hint)– Based on CPU MF counter data
• Captured by partition• Intended to eventually provide a more reliable way to choose a workload• Not yet validated – likely to be modified based on feedback
– Captured in EDF using CP3KEXTR* (one file per partition)– Not required
*Note: CP3KEXTR is a Load and Go program written in Assembler. Runs under z/OS.Available to customers since September 2010
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
190
System z New Workloads Sizing
� SAP
– Sizing for any IBM platform should be done through Techline (e.g., [email protected])
– Can also be used if you need a z114 sizing (less accurate) before announce (under Letter 112)
� zPSG– Sizing for new applications running on System z.
• Supports WebSphere (Application Server, Portal Server, Process Server, Message Broker and MQ), DB2 for z/OS Transaction sizing
– Engage Global Techline Solutions Sizing Support (choose your geography from the drop down)
• Additional support for workloads not found in zPSG• Cognos for Linux on System z• IBM Content Manager• Domino Mail
– Will zPSG be available to IBM BPs via Partner World
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
191
� Size the z196 portion or z114 portion– Engage a Techline specialist to help you collect the data and do the sizing via Deal Hub Connect– Use zPCR or zCP3000
• Use CP2KEXTR and CP3KVMXT to create an EDF file for z/OS and z/VM• Complete data collection guides located here:
– http://w3.ibm.com/techdocs/PRS2664 - for z/OS http://w3.ibm.com/techdocs/PRS2875 - for z/VM– IBM Business Partners will be able to obtain the tools directly from PartnerWorld refer to slide 24.
� Size the POWER7™ portion – allow at least one week.– Currently a manual process– 20-50 LPARS should take a week– More complex environments would take longer– Working towards a more automated process
� Sizing when migrating from competitive machines to POWER7 blades– Engage a Techline specialist via Deal Hub Connect to help you collect the data and do the sizing– Identify which machines and which time periods the customer would like to consider– Collect data from the competitive machines covering the time frames
• Server consolidation data collection guidance located here:• http://w3-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS1423
� Sizing new applications running on POWER7 blades– Engage Global Techline Solutions Sizing Support via Deal Hub Connect
• Software sizing questionnaires located here: • http://w3-03.ibm.com/support/techline/sizing/tg_im_sizing.html
Cross Platform sizing including POWER7 blades
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
192
Cross Platform sizing including POWER7 blades
� Sizing when the customer has an existing set of IBM servers they would like to
migrate to POWER7 blades (go ahead and collect the data now)– Identify which machines and which time periods the customer would like to consider in
the proposal – Collect data from AIX covering the time frames
• Work hand in hand with a POWER7 Specialist to collect the data and do the sizing
– OR• Engage a Techline specialist via Deal Hub Connect to help you collect the data and do the sizing • Sizing questionnaires located here: • http://w3-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS4034
� We're working with the WLE (Workload Estimator) team for support
– Targeted to be available at announce– Techline will deliver World Wide education when the support is available
� IBM Business Partners
– PWCS (PartnerWorld Contact Services)
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
193
Cross Platform sizing including Select IBM System x blades
� Size the z196 portion or z114 portion
– Engage a Techline specialist to help you collect the data and do the sizing via Deal Hub Connect
– Use zPCR or zCP3000• Use CP2KEXTR and CP3KVMXT to create an EDF file for z/OS and z/VM• Complete data collection guides located here:
– http://w3.ibm.com/techdocs/PRS2664 - for z/OShttp://w3.ibm.com/techdocs/PRS2875 - for z/VM
– IBM Business Partners will be able to obtain the tools directly from PartnerWorld.
� we're working with the WLE (Workload Estimator) team for support– Targeted to be available at announce– Techline will deliver World Wide education when the support is available
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
194
IBM Systems Workload Estimator
� Welcome to the IBM Systems Workload Estimator – IBMers access: http://wletest.rchland.ibm.com/wleProd/EstimatorServlet– IBM Business Partners access: http://www-912.ibm.com/wle/EstimatorServlet
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
195
IBM WebSphere DataPower XI50 for zEnterprise (DataPower XI50z)
� IBM
– DataPower XI50z can perform a tremendous variety of use cases, and in almost all real world use cases, there are other servers involved, which introduce their own latency.
– Currently DataPower XI50z sizing typically is performed on a per client basis, often based on results of a proof of concept.
– If a System z seller identifies the opportunity, they should bring in a SWG WebSphereseller who can then decide whether to bring in a SWG DataPower Tech Sales Specialist.
� IBM Business Partners– Must be certified to sell DataPower XI50z – Should already posses the skills to do the sizing
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
196
IBM zEnterprise 114 (z114)
Upgrades
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
197
z9 BC
z10 BC
M10
M05
z114z114
z196 Model M15
Yes
Air Cooled onlyNO
M10 upgrade path only
M05 upgrade Not Supported
� Upgrade paths from z9 BC and z10 BC
� Upgrade path to z196 Model M15 (Air cooled only)
� Disruptive upgrade M05 to M10 and from M10 to z196 M15
z114 Upgrade paths
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
198
z114
z890
z9 BC
z10 BC
zEnterprise 114 Product Positioning
� New modular based packaging construct.
� Increased capacity in a single footprint
– Designed for up to 1.12 times the z10 BC in total system capacity
– out-of-order instruction processing– higher clock frequency– larger caches– compiler enhancements
� New I/O subsystem infrastructure
� Connectivity improvements include bandwidth and throughput
� RAIM fault tolerant memory
� Federated capacity and hybrid computing capabilities via integration with zBX, application server blades, optimizers and Unified Resource Manager.
MIP
S c
ap
ac
ity i
n a
sin
gle
fo
otp
rin
t
PUs for Customer use
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
199
Summary
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
200
zEnterprise 114 in a nutshell:
� Integration with the zEnteprise BladeExtension
� Two Models
� Increased capacity in a single footprint
• 12s0 technology• out-of-order instruction processing• higher clock frequency• larger cache
� Robust Memory
� Upgrades
– Investment protection with upgrades from two previous families• z10 BC • z9 BC
– Upgradeability to z196 (M15)
� I/O Improvements– new I/O features and New I/O Drawer
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
201201
Extending the reach and strategic role of the mainframe across the enterprise; simplify and reduce the range of skills necessary for managing the datacenter;
break down cultural divides.
� Scale without adding complexity to meet the growing demands on the infrastructure
� Reduce cost through centralized data centermanagement and automation
� Fit for purpose application deployment
� Unify and optimize multiple architectures to work as a single system
� Improve performance with co-location of data and applications
� Improve resiliency through a private network
� Enable innovation through flexibility and breaking down silos
Summary:Deploy Multi-tier workloads on a single integrated system Create an integrated infrastructure that aligns to end-to-end business processes across technology boundaries
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
202
Questions ?
Parwez Hamid
Luiz Fadel
IBM zEnterprise™ System
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
203
Thank You
TakDanishDanke
German
Dank uDutch
ObrigadoBrazilian
Portuguese
������Thai
GrazieItalian
go raibh maith agatGaelic
TrugarezBreton
MerciFrench
GraciasSpanish
СпаСпаСпаСпаcибоибоибоибо
Russian
ந�றிTamil
ध�यवादHindi
���ًاArabic
감사합니다감사합니다감사합니다감사합니다Korean
תודה רבה Hebrew
Tack så mycketSwedish
DankonEsperanto
ありがとうございます
Japanese
谢谢谢谢谢谢谢谢Chinese
děkujiCzech
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
204
End of Presentation
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
205
I/O Cage/Drawer Migration
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
206
z114 & z196•4 I/O domains•32 I/O slots (PCIe I/O cards only)• At least two PCIe fanouts (4 ports per drawer)•7 EIA Units
z114, z196, & System z10 BC•2 I/O domains•8 I/O slots (legacy I/O cards only)• At least 2 HCA2-C fanouts (2 ports per drawer)•Up to two drawers on a pair of fanouts•5 EIA Units
z196 & earlier (except z10 BC)•7 I/O domains•28 I/O slots (legacy I/O cards only)•Up to 4 fanouts for all 7 domains•14 EIA Units
z114
Not Supported
I/O drawer
FC 4000
PCIe I/O drawer
FC 4003
z114 & z196 GA2 I/O Drawers and Cages
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
207
z196/z114 I/O Cage/Drawer Migration
� With the introduction of the PCIe I/O drawer, a strategy has been developed to help
migrate customer’s I/O from the current I/O cage and I/O drawer to the PCIe I/O drawer
� I/O cage/drawer migration goals
– Eliminate the need for I/O cages and I/O drawers by two generations after the z196
� I/O cage/drawer migration execution
– I/O feature ordering – I/O feature carry forward
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
208
I/O Feature Ordering and Carry Forward*
� Customers cannot order I/O cages (z196) or I/O drawers, they order I/O features and
eConfig will determine the proper mix of I/O cage (z196), I/O drawer, and the PCIe I/O drawer
– New builds, migration offerings, and z exchange programs • All FICON and OSA features (except 1000 BASE-T) will be PCIe I/O based• Crypto Express3, ESCON, ISC-3, OSA-Express3 1000BASE-T and PSC will be I/O
drawer based– Box MES and loose piece MES
• RPQ to order the I/O drawer based cards to fill empty I/O cage and I/O drawer slots until they are filled
– RPQ is only available if there are no open PCIe I/O drawer slots
– Current I/O drawer based cards are FICON Express8 and OSA-Express3, Crypto Express3 and ISC-3 coupling links
– Once the empty slots (i.e. I/O cage & I/O drawer) are filled, eConfig will drive out a PCIe I/O drawer with the associated features
� Customers will not be allowed to carry forward the oldest generation of I/O cards in future severs*
– z114 will not support (i.e. allow the carry forward) of FICON Express4, OSA-Express2 and ESCON features
– Future servers will not support FICON Express8, OSA-Express3, ISC-3, or PSC features
*Refer to the Statements of Direction
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
209
*All statements regarding IBM's plans, directions, and intent are subject to change or withdrawal without notice. Any reliance on
these statements of general direction is at the relying party's sole risk and will not create liability or obligation for IBM.
ESCON Statement of Direction* - February 15, 2011
� The IBM zEnterprise 196 (z196) will be the last high-end server to support ESCON
channels: IBM plans not to offer ESCON channels as an orderable feature on high-end System z servers which follow the z196 (machine type 2817). In addition, ESCON channels cannot be carried forward on an upgrade to such a follow-on server.
� This plan applies to channel path identifier (CHPID) types CNC, CTC, CVC, and CBY and to feature code numbers 2323 and 2324. System z customers should continue
migrating from ESCON to FICON. Alternate solutions are available for connectivity to ESCON devices.
� IBM Global Technology Services offers an ESCON to FICON Migration solution, This offering should help customers to simplify and manage a single physical and
operational environment.
� Notes: – This new Statement of Direction supersedes the previous ESCON SOD in Announcement letter 110-
170 of July 22, 2010. It also confirms the SOD in Announcement letter 109-230 of April 28, 2009 that “ESCON Channels will be phased out.”
– This SOD does NOT say that the z10 BC will be the last midrange server to support ESCON channels or the last to offer ESCON channels as an orderable feature.
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
210
IBMIBMIBMIBM
Old Current
Partnership
Prizm is available from IBM GTS Site & Facilities as part of the EFM Service (ESCON to FICON Migration - offering # 6948-97D)
http://www-935.ibm.com/services/us/index.wss/itservice/igs/a1026000?cm_re=masthead-_-itservices-_-site
The order process for Prizm is the same as it is for IBM cabling systems
FICON Convertor Support
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
211
Optica Technologies PRIZM FICON Converter http://www.opticatech.com/ To help reduce or eliminate ESCON channels on the host while maintaining ESCON and Bus/Tag-based devices and applications
What is PRIZM?� A purpose-built appliance designed exclusively for IBM System z to enable a smooth transition from use of
ESCON channels on the host to use of FICON channels on the host
� Allows ESCON devices to connect to FICON channels and FICON fabrics/networks – Also designed to support attachment of parallel (bus/tag) devices to FICON channels via ESBT
module
� Converts 1 or 2 FICON channels into 4, 8 or 12 ESCON channels– Replace aging ESCON Directors with PRIZM– Achieve streamlined infrastructure and reduced Total Cost of Ownership
� Qualified by the IBM Poughkeepsie, NY Vendor Solutions Lab in July 2009 for all ESCON devices– Refer to:
• http://www-03.ibm.com/systems/z/hardware/connectivity/index.html• Products -- > FICON / FCP Connectivity -- > Other supported devices
� PRIZM is available via IBM Global Technology Services: ESCON to FICON Migration offering (#6948-97D)
BackESCON Ports:
MT-RJ
FICON Ports: LC Duplex Front
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
212
Topologies supported by PRISM�Local: direct attached or switched�Remote: ISL (cascaded) or IP channel extended
FICON Disk
FICON
FICON
ESCON
FICON Tape
ESCON Tape
FICON
ESBT Module
B/T
ESCON Tape
Parallel (B/T) Printers
Prizm supportslocal OR remote FICON attachment via ISL or IP WAN
Where Does PRISM Fit in the Data Center?
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
213
Additional Material
213
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
214
PCHID Ranges
I/O Card Slot Drawer Type A02B A09B A16BA26B
(M05 only)
LG01I/O Drawer DCA 1
PCIe I/O Drawer 100 - 103 180 - 183 X X
LG02I/O Drawer 100 - 10F 180 - 18F 200 - 20F 280 - 28F
PCIe I/O Drawer 104 - 107 184 - 187 X X
LG03I/O Drawer 110 - 11F 190 - 19F 210 - 21F 290 - 29F
PCIe I/O Drawer 108 - 10B 188 - 18B X X
LG04I/O Drawer 120 - 12F 1A0 - 1AF 220 - 22F 2A0 - 2AF
PCIe I/O Drawer 10C - 10F 18C - 18F X X
LG05I/O Drawer 130 - 13F 1B0 - 1BF 230 - 23F 2B0 - 2BF
PCIe I/O Drawer PCIe Switch 0 X X
LG06I/O Drawer DCA 2
PCIe I/O Drawer 110 - 113 190 - 193 X X
LG07I/O Drawer 140 - 14F 1C0 - 1CF 240 - 24F 2C0 - 2CF
PCIe I/O Drawer 114 - 117 194 - 197 X X
LG08I/O Drawer 150 - 15F 1D0 - 1DF 250 - 25F 2D0 - 2DF
PCIe I/O Drawer 118 - 11B 198 - 19B X X
LG09I/O Drawer IFB MP
PCIe I/O Drawer 11C - 11F 19C - 19F X X
LG10I/O Drawer 160 - 16F 1E0 - 1EF 260 - 26F 2E0 - 2EF
PCIe I/O Drawer FSP-1,1 X X
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
215
PCHID Ranges
I/O Card Slot Drawer Type A02B A09B A16BA26B
(M05 only)
LG11I/O Drawer 170 - 17F 1F0 - 1FF 270 - 27F 2F0 - 2FF
PCIe I/O Drawer 120 - 123 1A0 - 1A3 X X
LG12I/O Drawer X X X X
PCIe I/O Drawer 124 - 127 1A4 - 1A7 X X
LG13I/O Drawer X X X X
PCIe I/O Drawer 128 - 12B 1A8 - 1AB X X
LG14I/O Drawer X X X X
PCIe I/O Drawer 12C - 12F 1AC - 1AF X X
LG15I/O Drawer X X X X
PCIe I/O Drawer PCIe Switch 2 X X
LG16I/O Drawer X X X X
PCIe I/O Drawer 130 - 133 1B0 - 1B3 X X
LG17I/O Drawer X X X X
PCIe I/O Drawer 134 - 137 1B4 - 1B7 X X
LG18I/O Drawer X X X X
PCIe I/O Drawer 138 - 13B 1B8 - 1BB X X
LG19I/O Drawer X X X X
PCIe I/O Drawer 13C - 13F 1BC - 1BF X X
LG20I/O Drawer X X X X
PCIe I/O Drawer 140 - 143 1C0 - 1C3 X X
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
216
PCHID Ranges
I/O Card Slot Drawer Type A02B A09B A16BA26B
(M05 only)
LG21I/O Drawer X X X X
PCIe I/O Drawer 144 - 147 1C4 - 1C7 X X
LG22I/O Drawer X X X X
PCIe I/O Drawer 148 - 14B 1C8 - 1CB X X
LG23I/O Drawer X X X X
PCIe I/O Drawer 14C - 14F 1CC - 1CF X X
LG24I/O Drawer X X X X
PCIe I/O Drawer PCIe Switch 3 X X
LG25I/O Drawer X X X X
PCIe I/O Drawer 150 - 153 1D0 - 1D3 X X
LG26I/O Drawer X X X X
PCIe I/O Drawer 154 - 157 1D4 - 1D7 X X
LG27I/O Drawer X X X X
PCIe I/O Drawer 158 - 15B 1D8 - 1DB X X
LG28I/O Drawer X X X X
PCIe I/O Drawer 15C - 15F 1DC - 1DF X X
LG29I/O Drawer X X X X
PCIe I/O Drawer FSP-1,2 X X
LG30I/O Drawer X X X X
PCIe I/O Drawer 160 - 163 1E0 - 1E3 X X
© 2011 IBM Corporation
yourdotcomibm.com International Technical Support Organization and Authoring Services
217
PCHID Ranges
I/O Card Slot Drawer Type A02B A09B A16BA26B
(M05 only)
LG31I/O Drawer X X X X
PCIe I/O Drawer 164 - 167 1E4 - 1E7 X X
LG32I/O Drawer X X X X
PCIe I/O Drawer 168 - 16B 1E8 - 1EB X X
LG33I/O Drawer X X X X
PCIe I/O Drawer 16C - 16F 1EC - 1EF X X
LG34I/O Drawer X X X X
PCIe I/O Drawer PCIe Switch 1 X X
LG35I/O Drawer X X X X
PCIe I/O Drawer 170 - 173 1F0 - 1F3 X X
LG36I/O Drawer X X X X
PCIe I/O Drawer 174 - 177 1F4 - 1F7 X X
LG37I/O Drawer X X X X
PCIe I/O Drawer 178 - 17B 1F8 - 1FB X X
LG38I/O Drawer X X X X
PCIe I/O Drawer 17C - 17F 1FC - 1FF X X