1, the fill test data is larger than the physical memory, at least more than the Innodb_buffer_pool_size value, you can not load all the data into memory, unless you want to test the full memory state of MySQL performance.
2. After each round of testing, restart the Mysqld instance and delete the system cache with the following method, releasing swap (if Swap is used), or even restarting the entire OS.
[[Email protected]]# Sync--Flush dirty data to disk
[[Email protected]]# echo 3 >/proc/sys/vm/drop_caches--Clear OS Cache
[Email protected]]# swapoff-a &&swapon–a
3, the pressure measurement results in addition to TPS/TPMC indicators, but also should pay attention to the system load data during the pressure measurement, especially iops,iowait,SVCTM ,%util, number of I/o bytesper second (I/O throughput ), transaction response time (tpcc-mysql/sysbench printed in the Test record ). In addition, if I/O devices can provide equipment-level IOPS, read-write delay data, you should also pay attention.
Copyright NOTICE: This article for Bo Master original article, without Bo Master permission not reproduced.
Some points to note in the MySQL stress test