April 16, 2014

The relationship between Innodb Log checkpointing and dirty Buffer pool pages

This is a time-honored topic, and there’s no shortage of articles on the topic on this blog. I wanted to write a post trying to condense and clarify those posts, as it has taken me a while to really understand this relationship. Some basic facts Most of us know that writing into Innodb updates buffer […]

Shard-Query adds parallelism to queries

Preamble: On performance, workload and scalability: MySQL has always been focused on OLTP workloads. In fact, both Percona Server and MySQL 5.5.7rc have numerous performance improvements which benefit workloads that have high concurrency. Typical OLTP workloads feature numerous clients (perhaps hundreds or thousands) each reading and writing small chunks of data. The recent improvements to […]

InnoDB thread concurrency

InnoDB has a mechanism to regulate count of threads working inside InnoDB. innodb_thread_concurrency is variable which set this count, and there are two friendly variables innodb_thread_sleep_delay and innodb_concurrency_tickets. I’ll try to explain how it works. MySQL has pluginable architecture which divides work between mysql common code (parser, optimizer) and storage engine. From storage engine’s point […]

InnoDB page size

As maybe you know InnoDB uses hard page size 16Kb for datafiles and for buffer pool. However this size can be changed if you need it for your workload. go to file innobase/include/univ.i, lines:

UNIV_PAGE_SIZE is page size (as you see – default value 16Kb). Possible values for UNIV_PAGE_SIZE is 8K, 16K, 32K, 64K. […]

Mess with innodb_thread_concurrency

In MySQL 5.0.19 the meaning of innodb_thread_concurrency variable was changed (yeah, again). Now innodb_thread_concurrency=0 means unlimitied count of concurrent threads inside InnoDB. It’s logical, but there was long way. In MySQL versions below 5.0.8 for unlimited threads you had to set innodb_thread_concurrency over 500 (and default value for innodb_thread_concurrency was 8 ). Starting with MySQL […]