IOPS QPS TPS

Yang Chilong:

http://blog.itpub.net/22664653/viewspace-767265/

http://blog.itpub.net/22664653/viewspace-767266/

This is not allowed:

Http://blog.chinaunix.net/uid-10661836-id-4472408.html

IOPS: (Input / Output operations Per Second, which is the number of I / O requests processed per second)
IOPS refers to how many times the storage can accept access from the host per second. A host's IO requires multiple accesses to the storage to complete. Here, the disk read and write capabilities are mentioned, such as it reads 100M per second and writes 50M. This description is Data throughput, while IOPS refers to the number of I / O requests per second. In detail, the number of requests is that reading 80M files is an I / O request, and writing 1K data is also an I / O request. Then the higher the IOPS value, the more corresponding requests can be accepted in a certain period of time. If you think about it, you will find that this is just a theory. Because the same request for reading 80M and writing 1K is naturally different. In addition to seeking and data transmission, there are many factors to consider, so if the IOPS is high enough, it will be more suitable for OLTP systems. For how to obtain the IOPS value, there are many tools on Linux and Windows. For testing, but the value of reference may not be much. If you want to improve IOPS, traditional solutions still use RAID stripe to improve I / O capabilities. In recent years, solid-state hard disk SSDs are very hot, and the technical indicators between different manufacturers are not It's the same.As for abnormal IOPS like Fusion-IO, you can do it to millions of levels. In general, using SSD can basically meet the needs. The performance of multiple SSD strips is still fierce. There is more to life.
IOPS calculation formula IOPS = 1000ms / (seek time + rotation delay time)
 




QPS (Query Per Second)
After talking about IOPS, talk about the very important QPS in the database. This indicator is available in all databases, but MySQL should pay more attention. It is easy to get the value of this indicator by executing the status command in MySQL. But this The value is a global indicator in the MySQL life cycle, but our system is not busy at all times, so what is the QPS at the peak of the system, we can only do it by ourselves. When we execute status, there is a Questions Although it is also a global indicator. But we can query this value every one second and subtract the two adjacent values to get the exact number of actual requests per second. If MySQL is busy, Then the value we get can be regarded as the peak response capability of MySQL QPS.
QPS calculation formula: Questions / Uptime (Uptime is replaced by a self-defined time unit)

mysql> show global status like "Questions"; 
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| Questions     | 10    |
+---------------+-------+
1 row in set (0.02 sec)

mysql> show global status like "Uptime";
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| Uptime        | 308   |
+---------------+-------+
1 row in set (0.02 sec)

TPS (Transcantion Per Second)
As for TPS .. It is also an important indicator for measuring the database. However, not every storage engine supports transactions. So take InnoDB as an example. TPS mainly involves commits and rollbacks.
TPS = (Commit + Rollback) / Seconds

mysql> show global status like "Com_commit";
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| Com_commit    | 0     |
+---------------+-------+
1 row in set (0.02 sec)

mysql> show global status like "Com_rollback";
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| Com_rollback  | 0     |
+---------------+-------+
1 row in set (0.01 sec)

IOPS QPS TPS