Category: Performance
-
MariaDB 10.1 can do 1 million queries per second
MariaDB 10.1 not only contains tons of new features, it has also been polished to deliver top performance. The biggest improvement has been achieved for scalability on massively multithreaded hardware. The following numbers show the throughput for a simplified sysbench OLTP benchmark on MariaDB-10.1.8 compared to MariaDB-10.0.21:
Benchmark Details
The benchmark is sysbench-mariadb (sysbench trunk with a fix for a more scalable random number generator) OLTP simplified to do 1000 point selects per transaction. The data set is 1 million […]
-
A few interesting findings on MariaDB and MySQL scalability, multi-table OLTP RO
t’s been almost a year since I benchmarked MariaDB and MySQL on our good old 4 CPU / 32 Cores / 64 Threads Sandy Bridge server. There seem to be a few interesting things happened since that time.
- MySQL 5.6.23 peak throughput dropped by ~8% compared to 5.6.14. Looks like this regression appeared in MySQL 5.6.21.
- 10.0.18 (git snapshot) peak threads increased by ~20% compared to 10.0.9 and reached parity with 5.6.23 (not with 5.6.20 though).
- 10.1.4 (git snapshot) and 5.7.5 are the champions (though 10.1.4 was usually 1-5% faster). Both have similar peaks @ 64 threads. They seem to be pushing this system limits, since both have capacity to scale more.
- 5.7.6 has serious (~50%) scalability regression compared to 5.7.5. There is heavy LOCK_plugin mutex contention, which seem to be caused by query rewrite plugin framework. I couldn’t find way to disable it. Hope this will be fixed in coming releases.
-
A few interesting findings on MariaDB and MySQL scalability, multi-table OLTP RO
It’s been almost a year since I benchmarked MariaDB and MySQL on our good old 4 CPU / 32 Cores / 64 Threads Sandy Bridge server. There seem to be a few interesting things happened since that time.
- MySQL 5.6.23 peak throughput dropped by ~8% compared to 5.6.14. Looks like this regression appeared in MySQL 5.6.21.
- 10.0.18 (git snapshot) peak threads increased by ~20% compared to 10.0.9 and reached parity with 5.6.23 (not with 5.6.20 though).
- 10.1.4 (git snapshot) and 5.7.5 are the champions (though 10.1.4 was usually 1-5% faster). Both have similar peaks @ 64 threads. They seem to be pushing this system limits, since both have capacity to scale more.
- 5.7.6 has serious (~50%) scalability regression compared to 5.7.5. There is heavy LOCK_plugin mutex contention, which seem to be caused by query rewrite plugin framework. I couldn’t find way to disable it. Hope this will be fixed in coming releases.
-
Performance evaluation of MariaDB 10.1 and MySQL 5.7.4-labs-tplc
Introduction
Evaluating the performance of database systems is a very demanding task. There are a lot of hard choices to be made, e.g.:
- What operating system and operating system version is to be used
- What configuration setup is to be used
- What benchmarks are to be used and how long are the warm-up and measure times
- What test setups are to be used
- What version of the database management system is used
- What storage engine is used
-
Lua sysbench – crash course
This is a follow-up on my previous blog post about using Lua enabled sysbench. Today I will dive into how to write Lua scripts for sysbench. Look at this simple example:
There are 3 functions prepare(), cleanup() and help(). Those are executed when sysbench is asked to perform the respective action (i.e. sysbench … help).
If you do sysbench … run then things are different. First sysbench starts the requested number of threads and executes the thread_init() function for each. […]
-
Using Lua-enabled sysbench
A quite common benchmark for MySQL is sysbench. It was written nearly 10 years ago by Alexey Kopytov.
Sysbench has modes to benchmark raw CPU performance, mutex speed, scheduler overhead and file IO performance. The probably most often used sysbench mode is OLTP. This benchmark mimics a OLTP scenario with small transactions hitting an optimized database. There are many variables to play with, most important is the number of simulated application threads (option –num-threads). The OLTP benchmark can be run […]
-
On performance of JDBC drivers.
Back when the first version of the MariaDB Java Client was released, someone asked in the comments about the performance characteristics of the driver compared to ConnectorJ. I answered with hand-waving, saying that nobody does anything stupid, the performance of the drivers would be roughly the same, but I promised to measure it and tell the world one day. And now that day has come. The day where three MySQL JDBC drivers (ConnectorJ, MariaDB JDBC, and Drizzle JDBC) are compared […]
-
MariaDB Introduces Atomic Writes
When dealing with high performance, low latency storage devices, such as SSD cards, one finds bottlenecks in new places. This is a story about such a bottle neck and how to work around it.
One unique feature of InnoDB is the double write buffer. This buffer was implemented to recover from half-written pages. This can happen in case of a power failure while InnoDB is writing a page (16KB = 32 sectors) to disk. On reading that page, InnoDB would […]