Scalablity and benchmark in mysql performance
- 1. Insert Picture Here Scalablity and Benchmark in MySQL performance Amrendra Kumar Principlal Member of Tech Staff
- 2. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.2 The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, and timing of any features or functionality described for Oracle’s products remains at the sole discretion of Oracle.
- 3. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.4 Why MySQL Performance ?.. ● Any solution may look “good enough”...
- 4. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.5 Why MySQL Performance ?.. ● Until it did not reach its limit..
- 5. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.6 Why MySQL Performance ?.. ● And reach a similar limit..
- 6. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.7 Why MySQL Performance ?.. ● And even improved solution may not resist to increasing load..
- 7. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.8 Why MySQL Performance ?.. ● Analyzing your workload performance and testing your limits may help you to understand ahead the resistance of your solution to incoming potentialproblems ;-)
- 8. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.9 The MySQL Performance Best Practice #1 is... ???.. USE Common Sense!!!... ;-) ● Any benchmark is fine but you must test it in your h/w and environment ● Test to believe
- 9. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.10 The MySQL Performance Best Practice #1 is... ???.. USE Common Sense!!!... ;-) Main Slides now :)
- 10. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.11 #2 - Monitoring is THE MUST ! � even don’t start to touch anything � without monitoring.. ;-)
- 11. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.12 MySQL Enterprise Monitor • Fantastic tool! • Did you already try it?.. Did you see it live?..
- 12. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.13 Other Monitoring Tools Cacti, Zabbix, Nagios, Solarwinds, etc….. • dim_STAT • well, I'm using this one, sorry ;-) • all graphs within presentation were made with it
- 13. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.14 A Word about Monitoring… always validate the impact of your Monitoring on your Production ;-) • taking 1sec measurements is fine, except : • if it’s eating 100% CPU time on one or more CPU cores.. • reducing your network traffic / latency.. • eats your RAM, etc. • avoid to be too much intrusive on MySQL/InnoDB internals.. • you may easily create an additional overhead • as well you may add artificial locks on your workflow • for ex: run in loop “show processlist”, etc.. • well, nothing is coming for free, so think about what you’re doing ! • (#1 best practice once again ;-))
- 14. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.15 • Single MySQL Instance Performance & Scalability • single HW host • no replication • just to understand how far MySQL Server may scale.. • what are the limits • what to care about ahead • which situations are absolutely to avoid.
- 15. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.16 Why Scalability ?.. • will x2 times faster CPU increase your performance by x2 ?.. • CPU cores : more and more over year-to-year.. • Intel 2CPU : 36cores-HT (2015) • intel 2CPU : 44cores-HT (March 2016).. • … • Scalability In Few Words : • your software is able to deliver a higher throughput if more HW resources are available.. • (then, scaling it well or not is another story ;-))
- 16. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.17 interpretation of Scalability… Scale upto N connections Both are scaling up to 64 connections, but only one is able to keep a higher load..
- 17. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.18 “Generic” Test Workloads @MySQL • Sysbench <= “Entry Ticket” • OLTP, RO/RW, N-tables, lots test workload load options, deadlocks • DBT2 / TPCC-like • OLTP, RW, very complex, growing db, no options, deadlocks • In fact using mostly only 2 tables! (thanks Performance Schema ;-)) • dbSTRESS • OLTP, RO/RW, several tables, one most hot, configurable, no deadlocks • iiBench • pure INSERT (time series) + SELECT • LinkBench (Facebook) • OLTP, RW, very intensive, IO-hungry.. • DBT3 • DWH, RO, complex heavy query, loved by Optimizer Team ;-)
- 18. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.19 RO Point-Selects @MySQL 5.7 (Apr.2016) • 1.8M QPS Sysbench Point-Selects 8-tab, 72cores-HT : • or even more, if you really run after numbers.. ;-))
- 19. Copyright © 2014, Oracle and/or its affiliates. All rights reserved.20 Q & A now... Reference:https://www.percona.com/live/data-performance-conference-2016/sessions/mysql-performance-scalability-benchmarks
Editor's Notes
- #2: &lt;number&gt;
- #20: point-select is representing a single SQL query reading a row by its primary key (PK) a workload based on such queries is very similar to any generic key-value store solution, but via SQL usually a point-select is extremely fast and doing a round-trip from end-to-end in database engine so any internals overhead in engine code, any contentions, any scalability limits are seen very quickly.. historically such kind of workload did not scale very well on MySQL/InnoDB tandem.. on MySQL Server side there was a huge amount of internal locks around every SQL query execution while on the InnoDB side there was a huge general overhead related to transactions management + related locks.. the first step ahead was made here in MySQL 5.6 but the real changes came only with MySQL 5.7 ;-)
- #21: https://www.percona.com/live/data-performance-conference-2016/sessions/mysql-performance-scalability-benchmarks