IBM Storage DS8700/DS8800 R6.1 Performance

IBM Systems & Technology Group
IBM System Storage DS8870 Release R7.3
Performance Update
Enterprise Storage Performance
Yan Xu
© 2014 IBM Corporation
IBM Systems & Technology Group
Agenda
2

Summary of DS8870 Hardware Changes

I/O Performance of High Performance Flash Enclosure (HPFE)

Easy Tier with HPFE

Copy Services with HPFE

New drives – 200GB/800GB SSDs

Disk Magic Support
© 2014 IBM Corporation
IBM Systems & Technology Group
R7.3 Hardware
Changes
3
© 2014 IBM Corporation
IBM Systems & Technology Group
R7.3 Hardware Changes

High Performance Flash Enclosure (HPFE) (performance and efficiency)

New enclosure that houses new high-performance flash adapters and Flash
Cards
– 1U storage enclosure.
– Contains up to 30 400GB Flash Cards (1.8” form factor)

Supported Configurations
– “All-Flash” HPFE system consists of only an A rack with up to 8 HPFEs.
•
Increased number of I/O bays (eight)
•
Flash Accelerator will be enabled by default for 16-core and p7+ configuration
– Hybrid supports up to 4 HPFEs in A rack besides the normal support for HDDs and/or SSDs.
– Only support for RAID5 in R7.3
4
© 2014 IBM Corporation
IBM Systems & Technology Group
Performance Benchmarks
•
OLTP Database Simulation Benchmarks
•
•
Sequential IO
•
•
Simulates IO done by OLTP applications. Similar to SPC-1. Do a mix of random reads and writes
with cache hits and cache misses.
• DBO(Database Open): 70% read/30% write, 4KB IO’s, 50% cache hit
• DB zOS(Database System Z): 75% read/25% write, 4KB IO’s, 72% cache hit
Large block sequential reads or writes to the storage server’s drives. Similar to SPC-2.
Corner benchmarks
•
•
•
•
Perform just one type of IO pattern. Most applications combine a mix of different IO patterns.
Cache Hits: 4KB random reads or writes to the storage server’s cache.
Cache Read Misses: 4KB random reads to the storage server’s drives.
Writes Miss: 4KB random writes to the storage server’s drives.
All results are from a DS8870 p7+, 16-core unless otherwise noted
5
© 2014 IBM Corporation
IBM Systems & Technology Group
HPFE I/O
Performance
6
© 2014 IBM Corporation
IBM Systems & Technology Group
HPFE vs SSD – Single Array
Single Array – RAID5 (6+p)
4KB Random
Sequential
100
1.5
60
1.8X faster
40
1
20
0.5
0
0
Read
Write
SSD
7
GB/s
KIO/s
80
2
1.4X faster
HPFE
Read
Write
SSD
HPFE
© 2014 IBM Corporation
IBM Systems & Technology Group
HPFE vs SSD – Single Array
4KB Random Read
Single Array, RAID5 (6+p)
Response Time (ms)
2.5
2.0
1.5
1.0
0.5
0.0
0
20
40
60
80
100
I/O Rate (KIOPS)
SSD
8
HPFE
© 2014 IBM Corporation
IBM Systems & Technology Group
HPFE vs SSD – 4 Arrays
Single HPFE/Single Device Adapter Pair
4 Arrays – RAID5
4KB Random
300
Sequential
4
2X faster
3
200
4X faster
150
100
2
50
1
0
0
Read
Write
SSD
9
GB/s
KIO/s
250
HPFE
Read
Write
SSD
HPFE
© 2014 IBM Corporation
IBM Systems & Technology Group
HPFE vs SSDs - DBO(70/30/50)
Single HPFE provides 3.2X IOPs improvement for OLTP (Database Open) workload
DBO workload: 70% reads/30% writes with 50% read cache hits.
2
Response Time (ms)
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
100
200
300
400
500
K IOps
1 HPFE
10
1 DA Pair w/ 32 SSDs
© 2014 IBM Corporation
IBM Systems & Technology Group
HPFE vs HDDs - DBO(70/30/50)
Four HPFE can achieve OLTP (DBO) IOPs provided by 1536 HDDs, with much
better response time.
DBO workload: 70% reads/30% writes with 50% read cache hits.
Response Time (ms)
10
8
6
4
2
0
0
100
200
300
400
500
600
700
K IOps
4 HPFEs
11
1536x 15K HDDs RAID10
© 2014 IBM Corporation
IBM Systems & Technology Group
Hybrid DS8870: HDD/HPFE
Note: Data allocation to HPFE was done manually
12
© 2014 IBM Corporation
IBM Systems & Technology Group
Hybrid DS8870: HDD/HPFE
Note: Data allocation to HPFE was done manually
13
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier with
HPFE
14
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: 15KRPM with SSDs or HPFE

Workload


15
DB2 Brokerage Transactional Workload
Configuration

DS8870, 256GB Cache, 8-core

144 300GB 15K drives – RAID-5

16 400GB SSDs – RAID-5 or 16 400GB HPFE Flash Cards – RAID-5

DB2 9.7 FP1, 4 Instances, 4 x 2 TB DBs, 4 Buffer Pools at 54 GB each
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: 15KRPM with SSDs or HPFE
DB2 Brokerage Transactional Workload
DB2 Overall Transaction Rate per Second
DB2 Brokerage Transactional Workload
16
14000
12000
3.5X
10000
8000
6000
4000
2000
2.8X
1.2X
15K RPM HDDs/SSDs,
with EasyTier
15K RPM HDDs/HPFE,
with EasyTier
0
15K RPM HDDs
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: 15KRPM with SSDs or HPFE
DB2 Brokerage Transactional Workload
DB2 response time (ms) of key trade activities
Market
Analysis
Customer
Position
Lookup
Trade
Trade
Order
Trade
Status
Trade
Update
Trade
Results
Overall
RT
15K RPM HDDs
7.65
18.31
2005.49
50.99
52.08
2794.97
63.39
166.91
15K RPM HDDS/HPFE,
with Easy Tier
2.42
13.71
503.21
37.23
29.16
805.38
45.53
48.20
Reduction (%)
68%
25%
75%
27%
44%
71%
28%
71%
17
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: 15KRPM with SSDs or HPFE
DB2 Brokerage Transactional Workload
DS8870 - Average Throughput/Response Time
Throughput (IOPs)
120000
15
100000
80000
10
60000
40000
5
20000
0
0
15K RPM HDDs
15K RPM HDDs/SSDs,
with EasyTier
Throughput
18
Volume Response Time (ms)
20
140000
15K RPM HDDs/HPFE,
with EasyTier
Volume Response time
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: 15KRPM with SSDs or HPFE

Workload


19
OLTP Workload with 40% read, 60% write
Configuration

DS8870, 256GB Cache, 8-core

192 300GB 15K drives – RAID-5

16 400GB SSDs – RAID-5 or 16 400GB HPFE Flash Cards – RAID-5
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: 15KRPM with SSDs or HPFE – OLTP Workload
OLTP
Response Time (ms)
10.0
8.0
6.0
7X vs HDD only
1.7X vs HDD/SSD
@3ms RT
4X
@3ms RT
4.0
2.0
0.0
0
50
100
150
200
250
I/O Rate (KIOPS)
192 300GB/15K(RAID-5)
192 300GB/15K(RAID-5) + 16 400GB SSDs(RAID-5), with Easy Tier
192 300GB/15K(RAID-5) + 16 400GB HPFE Flash Cards(RAID-5), with Easy Tier
20
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: intra-tier auto-rebalance SSDs and HPFE

Workload


21
OLTP Workload with 40% read, 60% write
Configuration

DS8870, 256GB Cache, 8-core

16 400GB SSDs – RAID-5

16 400GB HPFE Flash Cards – RAID-5
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: intra-tier auto-rebalance SSDs and HPFE
22
© 2014 IBM Corporation
IBM Systems & Technology Group
Easy Tier: intra-tier auto-rebalance SSDs and HPFE
OLTP
Response Time (ms)
10.0
8.0
6.0
4.0
> 40%
2.0
0.0
0
10
20
30
40
50
60
I/O Rate (KIOPS)
70
80
90
100
16 400GB HPFE Flash Cards, 16 400GB SSDs, RAID-5, initially skewed
16 400GB HPFE Flash Cards, 16 400GB SSDs, RAID-5, after auto-rebalance
23
© 2014 IBM Corporation
IBM Systems & Technology Group
Copy Services
with HPFE
24
© 2014 IBM Corporation
IBM Systems & Technology Group
Flashcopy Background Copy (without IO)
The Flashcopy background copy rate for DS8870 with HPFE showed near linear
scaling from 1 DA Pair (1HPFE) to 8 DA Pair (8 HPFE) configurations, with +70%
improvement comparing to DS8870 with HDDs
Background Copy
14
12
10
GB/s
8
6
4
2
0
1 DA-Pair
2 DA-Pairs
DS8870 w/HPFE
4 DA-Pairs
8 DA-Pairs
DS8870 w/HDD
HPFE: High Performance Flash Enclosure
25
© 2014 IBM Corporation
IBM Systems & Technology Group
Flashcopy with No-copy option: Database Open Workload
The Database Open (DBO) workload at equivalent throughput with both configurations. The
DS8870 with HPFE provides higher throughput and much lower response time (red text) with
either normal Flashcopy or Space Efficient Flashcopy.
DS8870 P7 with HDDs: 8-DA Pair, 384 15K HDDs, RAID5
DS8870 P7+ with HPFEs: 8-HPFEs, 240 HPFE Flash Cards, RAID5
Throughput (KIOPs)
DBO
90
80
70
60
50
40
30
20
10
0
Response time
0.3 ms
3.5 ms
13.2 ms
No FlashCopy
Standard FlashCopy
DS8870 - 8 Bluehawks Throughput
26
0.3 ms
0.3 ms
17.33 ms
Space Efficient
FlashCopy
DS8870 - 8 DA Pairs Throughput
© 2014 IBM Corporation
IBM Systems & Technology Group
New Drives
27
© 2014 IBM Corporation
IBM Systems & Technology Group
New Drives Supported


28
200GB 2.5” SAS SSD
800GB 2.5” SAS SSD
© 2014 IBM Corporation
IBM Systems & Technology Group
Open – 200GB & 800GB 2.5” SAS SSD Performance
The 200GB and 800GB 2.5” SAS SSDs on the DS8870 have equivalent performance
to the 400GB 2.5” SAS SSDs.
Single Rank – RAID-5 6+p
Sequential
Random
1600
60
1400
50
1200
MB/s
KIO/s
40
30
800
600
20
29
1000
400
10
200
0
0
4KB Random Read
4KB Random Write
200GB SSDs
400GB SSDs
800GB SSDs
Sequential Read
200GB SSDs
800GB SSDs
Sequential Write
400GB SSDs
© 2014 IBM Corporation
IBM Systems & Technology Group
Open – 200GB & 800GB 2.5” SAS SSD Performance
The rebuild rate of the 200GB and 800GB 2.5” SAS SSDs on the DS8870 is equal or
better than the 400GB 2.5” SAS SSDs.
Rebuild Rate, Single Rank (RAID-5 6+p)
Rebuild Rate (MB/s)
200
150
100
50
0
No Workload
200GB SSDs
30
Moderate
Workload
800GB SSDs
Heavy Workload
400GB SSDs
© 2014 IBM Corporation
IBM Systems & Technology Group
Disk Magic
Support
31
© 2014 IBM Corporation
IBM Systems & Technology Group
Disk Magic Support for R7.3

Support for HPFE
Support for new 200GB/800GB 2.5” SSDs

Delivery


Version 9.15.0 was released on June 6th
Note: Consult your IBM representatives and/or IBM Business
partners for Disk Magic Capacity studies for your environment.
32
© 2014 IBM Corporation
IBM Systems & Technology Group
Trademarks and Disclaimers
IBM Corporation 1994-2014 All rights reserved.
References in this document to IBM products or services do not imply that IBM intends to make them available in every country.
Trademarks of International Business Machines Corporation in the United States, other countries, or both can be found on the World Wide Web at
http://www.ibm.com/legal/copytrade.shtml.
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.
Linux is a registered trademark of Linus Torvalds 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.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Java and all Java-based trademarks are trademarks of Sun Microsystems, Inc. in the United States, other countries, or both.
Other company, product, or service names may be trademarks or service marks of others.
Information is provided "AS IS" without warranty of any kind.
The customer examples described are presented as illustrations of how those customers have used IBM products and the results they may have achieved. Actual
environmental costs and performance characteristics may vary by customer.
Information concerning non-IBM products was obtained from a supplier of these products, published announcement material, or other publicly available sources and does
not constitute an endorsement of such products by IBM. Sources for non-IBM list prices and performance numbers are taken from publicly available information,
including vendor announcements and vendor worldwide homepages. IBM has not tested these products and cannot confirm the accuracy of performance, capability, or
any other claims related to non-IBM products. Questions on the capability of non-IBM products should be addressed to the supplier of those products.
All statements regarding IBM future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.
Some information addresses anticipated future capabilities. Such information is not intended as a definitive statement of a commitment to specific levels of performance,
function or delivery schedules with respect to any future products. Such commitments are only made in IBM product announcements. The information is presented here
to communicate IBM's current investment and development activities as a good faith effort to help with our customers' future planning.
Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput or performance 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 or performance improvements
equivalent to the ratios stated here.
Photographs shown may be engineering prototypes. Changes may be incorporated in production models.
33
© 2014 IBM Corporation
IBM Systems & Technology Group
Thank You
34
© 2014 IBM Corporation