Amdahl GS - 1 -

Monitoring The Service Level Monitoring The Service Level DeliveredDelivered

Simon Stevens

Director Amdahl Software Division

Amdahl GS - 2 -

AmdahlAmdahlSoftware Software

‘‘The power behind e-systems’The power behind e-systems’

Amdahl GS - 3 -

Everyday Reality ...Everyday Reality ...

WFAC/DI/DO

NSDB

LMOS

PAWS

MAXLC

CAS GW

ATCs

MACs

IMCs

NETWORKSSNA

DATAKITX.25

TNETMVNFNLIN

Maintenance

CPCs

LMDCs

HMCs

VTAM

MVNFNDN

TCIS

DDD

MARCH

FACS /DSF

TIRKS

COS MOS

IFAS

NLIN

ACCS

NPSN

RCMACs

Provisioning

ACCs

VTAM

PREMIS

NCMS

BOSS/ICRISDOE

ICABS

SAGE

ACD

CCVRS

OLP

PARIS

VTAM

Customer Contact

Ne tworkEle me nts

DELPHI

LMOS/FE

S ARTS

S CRUBBER

SEAS

SOP

SWITCH

CAT

NMA

MLT

Ne tworkEle me nts

NLIN

CCRS

CAROT

FOMS

PICS

LFACSOPS /INE

PCsLANsServers

PCs LANs Servers

PCs LANsServers

Major Business ProcessesHighly Interdependent &Complex Technology PosesA Significant Challenge ForAvailability

Amdahl GS - 4 -

New Architectures, New New Architectures, New ChallengesChallenges

1 Element99.85% Availability

= 13Hr 8Min Lost / Year

10 Elements99.85% Availability

= 130Hr 30Min Lost / Year= 5.5 Full Days Downtime5.5 Full Days Downtime

AA BB CC DD EE** ** ****

99.8599.8599.8599.85

99.8599.85 99.8599.8599.8599.85

Availability = Availability = ** ......

Amdahl GS - 5 -

e-business puts you on e-business puts you on the front pagethe front page

eBay Goes Down Again.

eBAY: Shutdown halts auctions.

eBay's system out; stock down.

eBay Systems Crash Again.

eBay Down Four Hours. Stock Down 10%.

eBay Auction Closed.

eBay Failure Prompts Hot-site Backup Consideration.eBay Faces Another Outage.

eBay: An Innovator Falls off the Leading Edge.

eBay Recovers from Fourth Outage in a Month.

eBay has become the poster child for e-commerce disaster.

eBay Brand Suffers – The cost of downtime.

Amdahl GS - 6 -

RevenueRevenue direct lossdirect loss compensatory compensatory

paymentspayments lost future revenueslost future revenues investment lossesinvestment losses

Other ExpensesOther Expenses - temporary employees, overtime costs, extra - temporary employees, overtime costs, extra shipping costs, travel expenses, etc.shipping costs, travel expenses, etc.

Damaged ReputationDamaged Reputation customerscustomers supplierssuppliers financial marketsfinancial markets banksbanks business partnersbusiness partners etc.etc.

Financial PerformanceFinancial Performance revenue recognitionrevenue recognition cash flowcash flow lost discounts (A/P)lost discounts (A/P) payment guaranteespayment guarantees credit ratingcredit rating stock pricestock price

ProductivityProductivity No. of employees No. of employees

impacted x time lost impacted x time lost to poor service x to poor service x burdened hourly rateburdened hourly rate

How do service issues How do service issues cost you money?cost you money?

Do you Do you really really know?know?

Amdahl GS - 8 -

End-UserBased

ComponentBased

MONITORING PERSPECTIVEMONITORING PERSPECTIVE

End-user service and availability measurement is based on the compilation of information provided by many discrete technical component monitors.

Measurement is based on agents or sampling mechanisms co-located with end-users in order to collect accurate service and availability information without regard to the technical infrastructure.

Component BasedComponent Based End-User BasedEnd-User Based

Measurement ApproachesMeasurement Approaches

UNREALISTIC

REALISTIC

Amdahl GS - 9 -

A Matter Of LanguageA Matter Of Language

– CPU Utilisation– I/Os per second– Network Latency– Paging Ops. / Sec.– Transaction Turn-Around

Time– Packet Collision Rate– Bandwidth Utilisation– TX / Sec.– Automatic Failover

– Available– Responsive

Component BasedComponent Based End-User BasedEnd-User Based

Amdahl GS - 10 -

Measurement Approaches Measurement Approaches ComparedCompared

Component-based Measurement Gives Detailed Information For IT Technicians

End User-based Measurement Gives IT Managers Information To Share With Users– Provide a Foundation for Informed Discussion– Help Overcome Distrust– Develop Support For Investment Dialogue

And Can Detect Failures In Unmeasured Components

Amdahl GS - 12 -

Invoked

Continuous

MO

NIT

OR

ING

AP

PR

OA

CH

MO

NIT

OR

ING

AP

PR

OA

CH

Measurement is continuous meaning no user interaction to cause availability and response information to be collected.

Measurement is invoked by end-users. A user must do something in order for availability and response metrics to be collected.

ContinuousContinuousOperationOperation

ContinuousContinuousOperationOperation

InvokedInvokedOperationOperationInvokedInvoked

OperationOperation

Operational ApproachesOperational Approaches

PROACTIVE

REACTIVE

Amdahl GS - 13 -

Operational Approaches - Operational Approaches - The Human FactorThe Human Factor

10:05 The System Is Slow

10:06 I’ll Do Some Paperwork, Maybe It’ll Be Better Later

10:50 It’s Still Slow

10:51 “Hello, Helpdesk? This System Has Been Slow For The Last 45 Minutes!”

OR

It’s Always Slow In The Late Afternoon

I’ll Do Something Else

Amdahl GS - 14 -

Case StudyCase Study

Amdahl GS - 15 -

Bell Atlantic Service Delivery IssuesBell Atlantic Service Delivery Issues

Application Availability Problems

Erratic and Slow End User Response-times

Diminished Credibility With End Users

Increasing End User Dissatisfaction

Financial Penalties: Public Service Commission

Amdahl GS - 16 -

Availability TrendsAvailability TrendsTop Mission Critical Applications Hours Unavailable/Incidents

0

50

100

150

200

250

Hours

Hours

Incidents

4Q 94 1H 95 2H 95 1H 96 2H 96 1Q 97 2Q 97 3Q 97 4Q 97

Amdahl GS - 17 -

Availability Trends - Recovery TimeAvailability Trends - Recovery Time

107

97

41

2925

3631

2732

4Q 94 1H 95 2H 95 1H 96 2H 96 1Q 97 2Q 97 3Q 97 4Q 97

0

20

40

60

80

100

120

Minutes/Outage

Top Applications Average Minutes Per Outage

Amdahl GS - 18 -

Response Time - Top Systems

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

2-96

4-96

6-96

8-96

10-9

6

12-9

6

2-97

4-97

6-97

8-97

Sec

on

ds

Actual

Objective

I/T had more time to be “Proactive” for I/T had more time to be “Proactive” for improving improving Response Time from an End User PerspectiveResponse Time from an End User Perspective

Amdahl GS - 21 -

Focus on “Tall Trees”Focus on “Tall Trees”

1

2

3

13:45 14:00 14:15 14:30

Average

? ??

?

““What Is Causing This?”What Is Causing This?”

Amdahl GS - 22 -

End-UserEnd-UserBasedBased

ComponentComponentBasedBased

InvokedInvoked

ContinuousContinuous

MONITORING PERSPECTIVEMONITORING PERSPECTIVE

MO

NIT

OR

ING

AP

PR

OA

CH

MO

NIT

OR

ING

AP

PR

OA

CH

Proactive &Proactive &RealisticRealistic

Proactive butProactive butUnrealisticUnrealistic

Reactive &Reactive &UnrealisticUnrealistic

Reactive &Reactive &RealisticRealistic

EnView EnView The Amdahl SolutionThe Amdahl Solution

Proactive, Realistic

Availability and Service

Management

Proactive, Realistic

Availability and Service

Management

Amdahl GS - 23 -

EnView - example outputEnView - example output

0

10

20

30

40

50

60

70

80

90

12AM 1AM 2AM 3AM 4AM 5AM 6AM 7AM 8AM 9AM 10AM 11AM 12PM 1PM 2PM 3PM 4PM 5PM 6PM 7PM 8PM 9PM 10PM 11PM

Ave

rag

e H

ourl

y R

esp

onse

Tim

e in S

eco

nd

s

0

10

20

30

40

50

60

70

80

90

12AM 1AM 2AM 3AM 4AM 5AM 6AM 7AM 8AM 9AM 10AM 11AM 12PM 1PM 2PM 3PM 4PM 5PM 6PM 7PM 8PM 9PM 10PM 11PM

Ave

rag

e H

ourl

y R

esp

onse

Tim

e in S

eco

nd

s

Amdahl GS - 24 -

Technical OverviewTechnical Overviewof response time of response time

monitoring solutionmonitoring solution

Amdahl GS - 25 -

Script

WriteScript

Capture& Encode

011010010101010101

CaptureGUI

API

Dedicated NT Workstation

Installed at User Locations

Simulates Activities of Users at that Location

Accesses the Same Applications via the Same Network

Reports End-user Response Time, Error Messages or TimeoutRobotRobot

RobotRobot

RobotRobot

ReporterReporter

MonitorsMonitors

EnView Data FlowEnView Data Flow

CollectorCollectorLog

Log

Log

SNMP Gateway

SNMP Gateway

Web AccessWeb AccessNew York

Dallas

San Jose

Client/Server Application

Character-based Application

Web-based Application

Alerts &Alarms

Status &Trending

Notification Escalation

Dedicated NT Workstation

Receives/Consolidates Response Time and Error Messages from all EnView Robots

Compares Response Times with Pre-Defined Objectives

Dedicated NT Workstation

Strategically located (war room, help desk, network control, operations, etc.)

Real-time Status Display

Red, Yellow, Green

Dedicated NT Server

Microsoft SQLServer 7.0(Customer Provided)

Standard Reports

Access for ad-hoc Reporting

Available from Amdahl

Amdahl GS - 26 -

Who uses this approach ?Who uses this approach ?

Amdahl GS - 27 -

Some EnView CustomersSome EnView Customers

Bell Atlantic

US West

AT&T

First Union Bank

Bear Sterns

Grumann Northrop

Blue Cross Blue Shield

Co-operative Bank - SMILE

Deutsche Telekom

New York Housing Authority

SkyChief

Merrill Lynch

Marks & Spencer plc

British Airways

Amdahl GS - 28 -

Implementation of EnView at Bell Implementation of EnView at Bell AtlanticAtlantic

25 applications monitored: incl. Billing, Inventory, Customer Service, Work-Force Administration, Facilities Management, Service Order Processing, Switch Equipment Configuration

Over 175 transactions monitored (5 to 15 transactions per application)

Several different network protocols (VTAM, Telnet, TN3270)

Hardware installed: 90 robots (at 50 locations), 50 monitors and 4 collectors

Amdahl GS - 29 -

What Do Customers Say What Do Customers Say About EnView?About EnView?

“EnView allows us to keep in touch with our end-users. Because of EnView, the end-to-end

availability of our business applications has been dramatically improved.”

Gerry HigginsVice PresidentBell Atlantic Corporation

Amdahl GS - 30 -

What Do Customers Say What Do Customers Say About EnView?About EnView?

“EnView continuously monitors key application availability and response, often identifying performance issues

before they impact our end-users.”

Tom FogartySenior Vice PresidentFirst Union Bank

Amdahl GS - 31 -

“IS organizations seeking to measure and improve service quality should evaluate

EnView.”

What does industry say What does industry say about EnView?about EnView?

“EnView checks the response times and warns the IT manager about any delays. It helps the manager know which areas of the network need attention.”

Abigail WarakerIT WeekJune 24, 1998

Donna ScottVice PresidentGartner Group

Amdahl GS - 32 -

SummarySummary

Amdahl GS - 33 -

EnView CharacteristicsEnView Characteristics

Real-time

End-to-end

Visible

Non-intrusive

Realistic

Proactive

End-UserEnd-UserBasedBased

ComponentComponentBasedBased

InvokedInvoked

ContinuousContinuous

MONITORING PERSPECTIVEMONITORING PERSPECTIVE

MO

NIT

OR

ING

AP

PR

OA

CH

MO

NIT

OR

ING

AP

PR

OA

CH

Proactive &Proactive &RealisticRealistic

Proactive butProactive butUnrealisticUnrealistic

Reactive &Reactive &UnrealisticUnrealistic

Reactive &Reactive &RealisticRealistic

Proactive, Realistic

Availability and Service

Management

Proactive, Realistic

Availability and Service

Management

Amdahl GS - 34 -

EnView BenefitsEnView Benefits Improves End User and Data Centre Productivity

Allows the Right Resources to Be Quickly Deployed in Response to Problems

Improves MIS Credibility With End Users

Improves End User Satisfaction

Points Out Where IT Management Can Get Quick Results If They Make Changes

Determines Impact of Changes

Provides the Data Needed to Accurately Set and Verify Service Level Agreements

Amdahl GS - 35 -

Collateral BenefitsCollateral Benefits

Displays the Impact of Change– New Application, Data or Environment,

Infrastructure, Acquisition, Merger

Ensures That All Applications Are Available Prior to Start of Day or After Disaster Recovery

Allows for Stress Testing New Applications and Disaster Recovery Plans

Amdahl GS - 36 -

Amdahl EnView.CommerceSM ServiceManagement Overview (2nd Mar 2000 - 8th Mar 2000)

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Average Reponse Time by Time of Day(Time-zone: GMT)

Re

sp

on

se

Tim

e i

n S

ec

on

ds

LUFTHANSABOOKING(Dublin) TRAVELOCITY(Dublin) EXPEDIA(Dublin)

Amdahl GS - 37 -

Questions?Questions?