Performance Baseline of Hitachi Data Systems UCP for Oracle

Transcription

1 Performance Baseline of Hitachi Data Systems UCP for Oracle Part IV: Benchware Performance Suite Release 8.4 (Build ) August 2013

2 Contents 1 Introduction to Tests 2 Database Configuration 3 Benchmark Results Database Load 4 Benchmark Results OLTP Transactions 5 Reviewing Database Benchmark Results copyright 2013 by benchware.ch slide 2

3 Why measure Database performance? Projects need understandable key performance metrics for capacity planning - Data load - Data scan - Data aggregation - OLTP transactions - Time windows for certain operations copyright 2013 by benchware.ch slide 3

4 What is measured? Speed of single thread - Rows per second [rps] - Transactions per second [tps] - SQL service time [s] Maximum throughput of system - Rows per second [rps] - Transactions per second [tps] - SQL service time [s] - REDO rate [MBps] - REDO service time [s] Scalability - Throughput per process for n = {1, 2, 4, 8,, n} Efficiency of - All platform layers copyright 2013 by benchware.ch slide 4

5 How is measured? Benchware developed a specific Oracle scenario for each test case copyright 2013 by benchware.ch slide 5

6 Overview of Database load performance tests with Test Codes Data Load Un-compressed Test Code for Data Load via buffer cache Test Code for Data Load direct Conventional data load (LGWR stress test) DBL Bulk load - DBL-21 - Data Load Compressed Test Code for Data Load direct compressed Bulk load, compress BASIC DBL-31 Bulk load, compress OLTP DBL-32 Bulk load, compress HCC query low DBL-33 Bulk load, compress HCC archive low DBL-34 copyright 2013 by benchware.ch slide 6

7 Overview of Database aggregation performance tests with Test Codes Data Aggregation Create unique b-tree index Create non-unique b-tree index Test Code for Data on default Storage DBA-11 DBA-12 Data Scan Test Code for Data on default Storage Test Code for Data in Flash Cache Test Code for Data in Cell Flash Cache Full table scan DBS-12 DBS-13 DBS-14 copyright 2013 by benchware.ch slide 7

8 Overview of Database OLTP performance tests with Test Codes OLTP Transactions 1 hit per transaction Test Code for Data on default Storage Test Code for Data in Flash Cache Test Code for Data in Cell Flash Cache Select transaction DBX-12 DBX-13 DBX-14 Update transaction DBX-22 DBX-23 DBX-24 copyright 2013 by benchware.ch slide 8

9 Remarks on other benchmark tools... SAP, TPC, Swingbench, Hammerora, - No specific database performance metrics for data scan data load data aggregation - No support for Oracle flash cache or Oracle cell flash cache - No support for Oracle compression techniques copyright 2013 by benchware.ch slide 9

10 Contents 1 Introduction to Tests 2 Database Configuration 3 Benchmark Results Database Load 4 Benchmark Results OLTP Transactions 5 Reviewing Database Benchmark Results copyright 2013 by benchware.ch slide 10

11 Database Configuration Database Oracle Release Partition Option Real Application Cluster Data Guard HDS UCP Yes No No Basic Configuration Block size [kbyte] Archiving Force Logging Flashback HDS UCP 8 Yes Yes No copyright 2013 by benchware.ch slide 11

12 Database Configuration Memory Configuration SGA target [GByte] PGA target [GByte] Keep pool [GByte] Recycle pool [GByte] Default pool [GByte] HDS UCP REDO Configuration REDO log file size [GByte] #REDO log files groups #REDO log file members HDS UCP copyright 2013 by benchware.ch slide 12

13 Contents 1 Introduction to Tests 2 Database Configuration 3 Benchmark Results Database Load 4 Benchmark Results OLTP Transactions 5 Reviewing Database Benchmark Results copyright 2013 by benchware.ch slide 13

14 Load rate in [rps] Database transactional load, single process, different transaction size 12'000 10'000 8'000 6'000 With transaction size = 1 rpt one process can execute up to commit per second With avg service time < 250 μs per SQL insert statement HDS UCP Flash 4'000 2' Transaction size in rows per transaction [rpt] copyright 2013 by benchware.ch slide 14

15 HDS UCP with Flash Database transactional load, single process, different transaction size TX CPU Throughput Throughput SQL service Physical Physical Physical REDO REDO REDO REDO REDO Elap size busy rows/sec txn/sec time write write write size writes svt sync sync svt time Run Tst Code #N #J #T [rpt] [%] [rps] [tps] [s] [iops] [bps] [MBps] [MBps] [iops] [ms] writes [us] [s] DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E Legend: #N number of RAC nodes #J number of jobs #T number of threads (PX) [rps] rows per second [tps] transactions per second [iops] i/o operations per second [s] time in seconds [ms] time in milli seconds [μs] time in micro seconds copyright 2013 by benchware.ch slide 15

16 Load rate in [rps] Database transactional load, 2 rows per transaction 120' '000 80'000 60'000 High transactional insert rate with small transaction size of 2 rows > rows per second > commit per second system is cpu bound not I/O bound! HDS UCP Flash 40'000 20' Number of processes copyright 2013 by benchware.ch slide 16

17 HDS UCP with Flash Database transactional load, 2 rows per transaction TX CPU Throughput Throughput SQL service Physical Physical Physical REDO REDO REDO REDO REDO Elap size busy rows/sec txn/sec time write write write size writes svt sync sync svt time Run Tst Code #N #J #T [rpt] [%] [rps] [tps] [s] [iops] [bps] [MBps] [MBps] [iops] [ms] writes [us] [s] DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E Legend: #N number of RAC nodes #J number of jobs #T number of threads (PX) [rps] rows per second [tps] transactions per second [iops] i/o operations per second [s] time in seconds [ms] time in milli seconds [μs] time in micro seconds copyright 2013 by benchware.ch slide 17

18 Load rate in [rps] Database bulk load 1'400'000 1'200'000 1'000'000 Extreme high throughput: > 1.1 Mio rows per second data load peak REDO rate > 600 MBps system is cpu bound not I/O bound! 800'000 HDS UCP Flash 600' ' ' Number of processes copyright 2013 by benchware.ch slide 18

19 HDS UCP with Flash Database bulk load TX CPU Throughput Throughput SQL service Physical Physical Physical REDO REDO REDO REDO REDO Elap size busy rows/sec txn/sec time write write write size writes svt sync sync svt time Run Tst Code #N #J #T [rpt] [%] [rps] [tps] [s] [iops] [bps] [MBps] [MBps] [iops] [ms] writes [us] [s] DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E DBL E E E Legend: #N number of RAC nodes #J number of jobs #T number of threads (PX) [rps] rows per second [tps] transactions per second [iops] i/o operations per second [s] time in seconds [ms] time in milli seconds [μs] time in micro seconds copyright 2013 by benchware.ch slide 19

20 Contents 1 Introduction to Tests 2 Database Configuration 3 Benchmark Results Database Load 4 Benchmark Results OLTP Transactions 5 Reviewing Database Benchmark Results copyright 2013 by benchware.ch slide 20

21 Load rate in [tps] Oracle OLTP select performance, 1 row per transaction 200' ' ' μs 741 μs Extreme high throughput and short service times: ~ look up tx < 350 μs system is CPU bound - not I/O bound! 140' μs 120' '000 80' μs HDS UCP Flash 60' μs 40'000 20' μs 174 μs 187 μs Number of processes copyright 2013 by benchware.ch slide 21

22 HDS UCP with Flash Oracle OLTP select performance, 1 row per transaction CPU Throughput Throughput SQL service Physical Physical REDO Hitrate Hitrate Physical Physical Elap busy rows/sec txn/sec time read write write db flash exa flash read write time Run Tst Code #N #J #T [%] [rps] [tps] [s] [iops] [iops] [iops] [%] [%] [MBps] [MBps] [s] DBX E E E DBX E E E DBX E E E DBX E E E DBX E E E DBX E E E DBX E E E DBX E E E Legend: #N number of RAC nodes #J number of jobs #T number of threads (PX) [rps] rows per second [tps] transactions per second [iops] i/o operations per second [s] time in seconds copyright 2013 by benchware.ch slide 22

23 Load rate in [tps] Oracle OLTP update performance, 1 row per transaction 70'000 60'000 50' μs 941 μs μs Extreme high throughput and short service times: > update tx < 1 ms peak REDO rate 564 MBps system is CPU bound - not I/O bound! 40'000 30' μs HDS UCP Flash 20'000 10' Number of processes copyright 2013 by benchware.ch slide 23

24 HDS UCP with Flash Oracle OLTP update performance, 1 row per transaction CPU Throughput Throughput SQL service Physical Physical REDO Hitrate Hitrate Physical Physical Elap busy rows/sec txn/sec time read write write db flash exa flash read write time Run Tst Code #N #J #T [%] [rps] [tps] [s] [iops] [iops] [iops] [%] [%] [MBps] [MBps] [s] DBX E E E DBX E E E DBX E E E DBX E E E DBX E E E DBX E E E DBX E E E DBX E E E Legend: #N number of RAC nodes #J number of jobs #T number of threads (PX) [rps] rows per second [tps] transactions per second [iops] i/o operations per second [s] time in seconds [ms] time in milli seconds [μs] time in micro seconds copyright 2013 by benchware.ch slide 24

25 Contents 1 Introduction to Tests 2 Database Configuration 3 Benchmark Results Database Load 4 Benchmark Results OLTP Transactions 5 Reviewing Database Benchmark Results copyright 2013 by benchware.ch slide 25

26 Benchmark Results Summary Data Load Un-Compressed Conventional OLTP insert [rps] Bulk load [rps] Metric [rps] [rps] HDS UCP With Server internal Flash OLTP Transactions 1 hit per transaction Select transaction Throughput Service time Update transaction Throughput Service Time Metric [tps] [ms] [tps] [ms] HDS UCP With Server internal Flash copyright 2013 by benchware.ch slide 26

27 Benchmark Results Reviewing This is new: all database load benchmark tests are cpu-bound instead of I/O-bound Database load and OLTP transactions - Extreme high throughput - Extreme low service times HDS UCP with Flash is excellent platform for all kind of OLTP and DWH operations copyright 2013 by benchware.ch slide 27

28 swiss precision in performance measurement