Performance analysis of the HP Serviceguard Storage Management Suite for Oracle Database

ManualsBrandsHP ManualsSoftwareHP Serviceguard Storage Management Suite Software

Performance analysis of the HP Serviceguard Storage

Management Suite for Oracle Database

Executive summary............................................................................................................................... 2

Introduction......................................................................................................................................... 2

Test methodology................................................................................................................................. 3

Test topology....................................................................................................................................... 4

Hardware ....................................................................................................................................... 4

Software ......................................................................................................................................... 5

Disk configuration ............................................................................................................................ 5

Database configuration..................................................................................................................... 5

Test results .......................................................................................................................................... 6

Test analysis summary ........................................................................................................................ 14

Recommendation ............................................................................................................................... 14

For more information.......................................................................................................................... 15

Summary of content (15 pages)

PAGE 1
Performance analysis of the HP Serviceguard Storage Management Suite for Oracle Database Executive summary............................................................................................................................... 2 Introduction......................................................................................................................................... 2 Test methodology...........................................................................................................
PAGE 2
Executive summary For HP-UX 11i, the HP Serviceguard Storage Management Suite is ideally suited to customers who need the highest levels of availability and superior Oracle® Real Application Clusters (RAC) database performance or who have an application that would benefit from a clustered file system for improved manageability.
PAGE 3
The most important factor of performance improvement is Oracle’s ODM. ODM is a standard API specified by Oracle for database I/O. When Oracle performs an I/O operation, it uses the vendor‘s ODM library if provided. ODM improves both performance and manageability of the file system.
PAGE 4
• db_file_sequential_read wait time, the time a user process is waiting for a single physical I/O call to return when reading buffers into the Oracle System Global Area (SGA). • logfile sync time, the time a user process is waiting for the log writer to finish flushing the log buffer to disk. This occurs when a user commits a transaction (a transaction is not considered committed until all of the redo to recover the transaction has been successfully written to disk).
PAGE 5
Software For the database servers we used the following software packages: HP-UX B.11.23.0512.034 Feature Enablement Patches for HP-UX 11i v2, December 2005 T2777BA HP Serviceguard Cluster File System for RAC plus any required patch listed in the release notes Oracle 9i R2 RAC 9.2.0.6 Database Server product Software used on the client side is irrelevant for this paper. Disk configuration From the EVA4000 we presented 4 LUNs to the SAN network.
PAGE 6
Test results 1. One node RAC • TPMs 35000 Raw one node TPMs 30000 30960 CFS one node TPMs 26178 TPMs 25000 20000 19511 17950 15000 12648 11559 10000 10 20 40 Batch Users (light, medium, heavy workload) • CPU Utilization (%idle) 85% 81.4% 80.7% 80% CPU idle % 75% 75.0% 70% 68.5% 65% Raw one node CPU idle % 60% CFS one node CPU idle % 58.0% 55% 54.
PAGE 7
• I/O throughput per second I/O throughput is calculated by adding read + writes I/O operations reported in the Oracle statspack. 6000 5555 5500 5000 Raw one node i/o rate I/O throughput /sec CFS one node i/o rate 4641 4500 4000 3884 3559 3500 3000 2500 2369 2142 2000 10 20 40 Batch Users (light, medium, heavy workload) • DB file sequential read (ms) 16 DBfile sequential read (ms) 14 12 14 Raw one node DB seq. read CFS one node DB seq.
PAGE 8
2. Two node RAC • TPMs 45000 42093 40000 37800 Raw two node TPMs TPMs 35000 CFS two node TPMs 30000 27959 25988 25000 20000 19644 17928 15000 20 40 80 Batch Users (light, medium, heavy workload) • CPU Utilization (%idle) 85% 80% 79.5% 77.3% CPU idle % 75% 70% 68.9% 65% 60% 55% 64.9% Raw two node CPU idle % CFS two node CPU idle % 53.5% 52.
PAGE 9
• I/O throughput per second The I/O throughput is cumulative, which means that we added the read + write operations of node A and node B. 8000 7500 7000 6981 Raw two node i/o rate CFS two node i/o rate 6546 I/O throughput/sec 6500 6000 5519 5500 5131 5000 4500 4000 3673 3500 3355 3000 20 40 80 Batch Users (light, medium, heavy workload) • DB file sequential read 25 23 21 DB file seuqential read 19 21 Raw two node DB seq. read CFS two node DB seq.
PAGE 10
• CFS vs. Raw total network packets/sec 16647 Raw two node Unicast pkts/sec 16000 CFS two node Unicast pkts/sec 14000 14528 CFS two node CFS DLPI traffic only Unicast pkts/sec 12000 11191 10535 10000 8000 7517 7161 6000 4000 3454 2000 1567 1073 0 20 40 80 Batch Users (light, medium, heavy workload) • CR block receive time (ms) Average CR block receive times are identical for CFS and Raw and as such we plotted only one set of results.
PAGE 11
3. Three node RAC • TPMs 50000 47047 Raw three node TPMs 45000 CFS three node TPMs 43121 TPMs 40000 35000 33413 31669 30000 25000 24310 22886 20000 30 60 120 Batch Users (light, medium, heavy workload) • CPU Utilization (%idle) 85% 81.0% 80% 77.6% 75% CPU idle % 72.0% 70% 65% 69.2% Raw three node CPU idle % 64.6% CFS three node CPU idle % 60% 56.
PAGE 12
• I/O throughput per second 9000 8500 8000 8125 Raw three node i/o rate CFS three node i/o rate I/O throughput /sec 7500 7456 7000 6544 6500 6246 6000 5500 5000 4581 4500 4298 4000 30 60 120 Batch Users (light, medium, heavy workload) • DB file sequential read 30 27 25 Raw three node DB seq. read 23 DB file sequential read (ms) CFS three node DB seq.
PAGE 13
• CFS vs. Raw total network packets/sec 19000 Raw three node Unicast pkts/sec 17000 CFS three node Unicast pkts/sec 19521 16455 CFS two node CFS DLPI traffic only 15000 13651 Unicast pkts/sec 13000 11557 11000 9509 9000 8338 7000 5000 4950 3000 2245 1636 1000 30 60 120 Batch Users (light, medium, heavy workload) • CR block receive time (ms) 0.8 average CRblock receive time (ms) 0.7 Average CR block receive time 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.
PAGE 14
Test analysis summary The results demonstrate clearly that CFS scales as well as Raw SLVM does in a RAC environment if the database is partitioned appropriately, i.e. partitioning of the I/O write intensive tables. Let’s first consider the one node RAC test scenario. There is no CFS or database interconnect network traffic and the storage does not cause contention on the I/O server side. Therefore the difference in the TPMs rate can only be explained by the added overhead of the CFS/ODM stack.
PAGE 15
For more information http://www.hp.com/go/Serviceguard © 2006 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. HP shall not be liable for technical or editorial errors or omissions contained herein.