Ken Cantrell / NetApp Mark Rogov / EMC · The material contained in this tutorial is copyrighted by...

Ken Cantrell / NetApp

Mark Rogov / EMC

July 30, 2015

© 2015 Storage Networking Industry Association. All Rights Reserved.

About The Speakers

2

Mark RogovEMC

Systems Engineer

@rogovmark

Dr. J MetzCisco

R&D Engineer

@drjmetz

Ken CantrellNetApp

Manager Perf Engineering

@kencantrelljr


SNIA Legal Notice

The material contained in this tutorial is copyrighted by the SNIA unless otherwise noted.

Member companies and individual members may use this material in presentations and literature under the following conditions:

Any slide or slides used must be reproduced in their entirety without modification

The SNIA must be acknowledged as the source of any material used in the body of any document containing material from these presentations.

This presentation is a project of the SNIA Education Committee.

Neither the author nor the presenter is an attorney and nothing in this presentation is intended to be, or should be construed as legal advice or an opinion of counsel. If you need legal advice or a legal opinion please contact your attorney.

The information presented herein represents the author's personal opinion and current understanding of the relevant issues involved. The author, the presenter, and the SNIA do not assume any responsibility or liability for damages arising out of any reliance on or use of this information.

NO WARRANTIES, EXPRESS OR IMPLIED. USE AT YOUR OWN RISK.

3


Storage Performance Benchmarking

4

SYSTEM UNDER

TEST

METRICS AND

TERMINOLOGY

FILE

COMPONENTS

BLOCK

COMPONENTS

WORKLOAD

DEFINITIONS



5

SYSTEM UNDER

TEST

METRICS AND

TERMINOLOGY

FILE

COMPONENTS

BLOCK

COMPONENTS

WORKLOAD

DEFINITIONS

TODAY FUTURE WEBCASTS

INTRO BLOCK FILE GRAPH


BENCHMARKSBALANCED PROFESSIONAL

APPROACH

BUSINESSRUN PRODUCTION

APPLICATIONS

BAKE-OFFSSINGLE OBSCURE METRIC

6

Three Great Buckets Of Testing

UN*X DD

IOMETER

IOZONE

VDBENCH

DATABASE

COMPUTE FARM

VDI

LOGIN VSI

VMMARK

SPEC SFS

SPC1


© 2015 Storage Networking Industry Association. All Rights Reserved. 7

Block / SAN Terminology

IO IOIOIOIOIOIO BLOCK / SAN

PROTOCOLFC/iSCSI/FCoE

Targ

et

Init

iato

r




IO IOIOIOIOIOIO

Targ

et

BLOCK / SAN


IO

OPERATIONAKA TYPE:

READ

WRITE

Init

iato

r



Block / SAN Operations

9

Source: http://www.t10.org/cgi-bin/ac.pl?t=f&f=sbc4r07c.pdf

Graphic: http://www.wordle.net/create




IO IOIOIOIOIOIO

Targ

et

BLOCK / SAN


IO

OPERATIONAKA TYPE:

READ

WRITE

Init

iato

r

IOPSCOUNT EVERY

IO OPERATION

PER SECOND

THROUGHPUTSAME AS IOPS



Where To Measure IOPS?

11

HOSTS

STORAGE

CONTROLLERDISK DRIVES




12

HOSTS

STORAGE


500

400

300

200

100

FRONT END

1500




13

HOSTS

STORAGE


500

400

300

200

1001000

1000

200

IDLE

15002200

FRONT ENDBACK END




IO IOIOIOIOIOIO

Targ

et

BLOCK / SAN


IO

OPERATIONAKA TYPE:

READ

WRITE

Init

iato

r

IOPSCOUNT EVERY

IO OPERATION

PER SECOND

MB/SPAYLOAD SUM

OF EVERY

OPERATION

PER SECOND

BANDWIDTHSAME AS MB/S




10004

TERABYTES

TB

10003

GIGABYTES

GB

MB/s Vs. MiB/s

15

MEGABYTES

MB 10002

10001

KILOBYTES

KB

10244

TEBIBYTES

TiB

15

10243

GIBIBYTES

GiB

10242MEBIBYTES

MiB

10241

KIBIBYTES

KiB

10%

7%

5%

2%

DECIMAL BINARY




IO IOIOIOIOIOIO

Targ

et

BLOCK / SAN


IO

OPERATIONAKA TYPE:

READ

WRITE

IO SIZESIZE OF DATA

PAYLOAD

IN AN IO

OPERATION

Init

iato

r

IOPSCOUNT EVERY

IO OPERATION

PER SECOND

MB/SPAYLOAD SUM

OF EVERY

OPERATION

PER SECOND





MB/s (or MiB/s), IO Size, And IOPS

17

0

50

100

150

200

250

300

350

400

450

500

4,096 8,192 16,384 32,768 65,536 131,072 262,144

MiB

/s

IOPS

MiB/s For Different IO Sizes

512B

1KiB

4KiB

8KiB

16KiB

MiB/s = 𝐼𝑂 𝑆𝑖𝑧𝑒 ∗ 𝐼𝑂𝑃𝑆KEEPING IOPS

CONSTANT,

LARGER I/O SIZES

PRODUCE

GREATER MiB/S

512B

1KiB

4KiB



MB/s (or MiB/s), IO Size, And IOPS

18

0

50

100

150

200

250

300

350

400

450

500

4,096 8,192 16,384 32,768 65,536 131,072 262,144

MiB

/s

IOPS

MiB/s For Different IO Sizes

512B

1KiB

4KiB

8KiB

16KiB

MiB/s = 𝐼𝑂 𝑆𝑖𝑧𝑒 ∗ 𝐼𝑂𝑃𝑆AS IO SIZE

INCREASES

IOPS DECREASE

FOR THE SAME

MiB/S

16KiB 8KiB 4KiB



IO SIZESIZE OF DATA

PAYLOAD

IN AN IO

OPERATION

19


IO IOIOIOIOIOIO

Targ

et

BLOCK / SAN


IO

OPERATIONAKA TYPE:

READ

WRITE

Init

iato

r

IOPSCOUNT EVERY

IO OPERATION

PER SECOND

MB/SPAYLOAD SUM

OF EVERY

OPERATION

PER SECOND


RESPONSE

TIMETIME TARGET

TAKES TO REPLY

TO AN IO

LATENCYSAME AS

RESPONSE

TIME




Aggregate Property Of Response Time

20

DISK DRIVES

STORAGE

CONTROLLER HOSTS




21

DISK DRIVES

STORAGE

CONTROLLER HOSTS

WRITE ≥ 1ms1ms




22

DISK DRIVES

STORAGE

CONTROLLER HOSTS

READ

WRITE

≥ 1.5ms

≥ 1ms1ms

1ms

0.5ms




23

DISK DRIVES

STORAGE

CONTROLLER HOSTS

READ

WRITE

≥ 1.5ms

≥ 1ms

READ

1ms

1ms

≥ 5ms

4ms

0.5ms



File / NAS Terminology

OP OPOPOPOPOPOP FILE

PROTOCOLNFS/SMB/FTP

Clie

nt

Serv

er




25


PROTOCOLNFS/SMB/FTP

Clie

nt

Serv

er

FILE

OPERATIONSEVERAL DOZEN

DEPENDING

ON

PROTOCOL



NFSv4 Operations

26

Source: https://www.ietf.org/rfc/rfc3530.txt




SMB3 Operations

27

Source: https://msdn.microsoft.com/en-us/library/cc246482.aspx(use [MS-SMB2].pdf)





28


PROTOCOLNFS/SMB/FTP

Clie

nt

Serv

er

FILE


DEPENDING

ON

PROTOCOL

OPSCOUNT EVERY

PROTOCOL

OPERATION

PER SECOND




29


PROTOCOLNFS/SMB/FTP

Clie

nt

Serv

er

FILE


DEPENDING

ON

PROTOCOL

OPSCOUNT EVERY

PROTOCOL

OPERATION

PER SECOND

MB/SPAYLOAD SUM

OF EVERY

OPERATION

PER SECOND

THROUGHPUTSAME AS MB/S





30


PROTOCOLNFS/SMB/FTP

Clie

nt

Serv

er

FILE


DEPENDING

ON

PROTOCOL

OPSCOUNT EVERY

PROTOCOL

OPERATION

PER SECOND

MB/SPAYLOAD SUM

OF EVERY

OPERATION

PER SECOND






31


PROTOCOLNFS/SMB/FTP

Clie

nt

Serv

er

FILE


DEPENDING

ON

PROTOCOL

OPSCOUNT EVERY

PROTOCOL

OPERATION

PER SECOND

MB/SPAYLOAD SUM

OF EVERY

OPERATION

PER SECOND



RESPONSE

TIMETIME TARGET

TAKES TO REPLY

TO AN IO

LATENCYSAME AS

RESPONSE

TIME



Difference Between OPS and MB/s (MiB/s)

Two Otherwise Identical Runs With Different NFS Operations Mix (64 KiB IO Size)

32

70

30

High MetaDataOperations

Stat Read

30

70

High Data Operations

Stat Read0

5

10

15

20

25

30

35

40

45

0

200

400

600

800

1000

1200

1400

1600

1800

2000

OPS Read MiB/s Stat MiB/s

High MetaData

Operations

High Data

Operations



Graph Fun Or Why “It Depends” Is Relevant?

33

0 2000 4000 6000 8000 10000 12000 14000

Operations per second, OPS

5,000 10,000




34

10

5

0

2

4

6

8

10

12

0 2000 4000 6000 8000 10000 12000 14000

Re

sp

onse

tim

e,

ms

Operations per second, OPS

Response time per OPS

system 1 system 2




35

0

5

10

15

20

25

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000


System 1 System 2

Operations per sec, OPS

Re

sp

onse

tim

e,

ms




36

0

5

10

15

20

25

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000


System 1 System 2


Re

sp

onse

tim

e,

ms



MAXIMUM

PERFORMANCE

NEED


37

0

5

10

15

20

25

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000


System 1 System 2


Re

sp

onse

tim

e,

ms



MINIMUM

PERFORMANCE

REQUIREMENT


38

0

5

10

15

20

25

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000


System 1 System 2


Re

sp

onse

tim

e,

ms



ACCEPTABLE LATENCY BAND


39

0

5

10

15

20

25

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000


System 1 System 2


Re

sp

onse

tim

e,

ms



ACCEPTABLE LATENCY BAND

Variance

40

0

5

10

15

20

25

10

00

12

50

15

00

17

50

20

00

22

50

25

00

27

50

30

00

32

50

35

00

37

50

40

00

42

50

45

00

47

50

50

00

52

50

55

00

57

50

60

00

62

50

65

00

67

50

70

00

72

50

75

00

77

50

80

00

82

50

85

00

87

50

90

00

92

50

95

00

97

50

10

000

10

250

10

500

10

750

11

000

11

250

11

500

11

750

12

000


System 1 System 2


Re

sp

onse

tim

e,

ms



Other Metrics To Consider

41

0

5

10

15

20

25

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000


System 1 System 2


Re

sp

onse

tim

e,

ms

$2/OP

$10/OP

$1.1/OP

$5.5/OP




42

0

5

10

15

20

25

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000


System 1 System 2


Re

sp

onse

tim

e,

ms

$1.1/OP

$5.5/OP

$25K/U

$1K/U

$40/W

$140/W

} @9000 OPS @ 17 ms

} @9000 OPS @ 10 ms




43

0

5

10

15

20

25

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000


System 1 System 2


Re

sp

onse

tim

e,

ms

$1.1/OP

$5.5/OP

$25K/U

$1K/U

$40/W

$140/W

} @9000 OPS @ 17 ms

} @9000 OPS @ 10 ms

CONTEXTMAKES

METRICSMATTER



Summary

44

System 1 System 2

GRAPH FUN

CONTEXTMAKES

METRICSMATTER

OPSCOUNT EVERY

PROTOCOL

OPERATION

PER SECOND

MB/SPAYLOAD SUM

OF EVERY

OPERATION

PER SECOND

TERMINOLOGY

IOPSCOUNT EVERY

IO OPERATION

PER SECOND

RESPONSE

TIMETIME TARGET

TAKES TO REPLY

TO AN IO




45

SYSTEM UNDER

TEST

METRICS AND

TERMINOLOGY

FILE

COMPONENTS

BLOCK

COMPONENTS

WORKLOAD

DEFINITIONS

TODAY FUTURE WEBCASTS


After This Webcast

46

This Webcast and a PDF of the slides will be posted to the SNIA Ethernet

Storage Forum (ESF) website and available on-demand

http://www.snia.org/forums/esf/knowledge/webcasts

A full Q&A from this webcast, including answers to questions we couldn't get

to today, will be posted to the SNIA-ESF blog

http://sniaesfblog.org/

Follow us on Twitter @SNIAESF

Next Webcast - October 21st

“Storage Performance Benchmarking: Part 2”

http://www.snia.org/forums/esf/knowledge/webcasts

http://sniaesfblog.org/

https://twitter.com/SNIAESF

Ken Cantrell / NetApp Mark Rogov / EMC · The material contained in this tutorial is copyrighted by...

Documents

Transcript of Ken Cantrell / NetApp Mark Rogov / EMC · The material contained in this tutorial is copyrighted by...