MySQL 5.6.10 vs Percona 5.6.6 TPCCC performance test

1. tpcc test results

Click to view the chart

2. Test Environment

3. Key configuration options of InnoDB

Note:
1. ring replication: it refers to the creation of ring replication on three nodes, each of which is master/slave. During the test, we also tried to simulate master downtime, and slave switched the master randomly to verify the data reliability in GTID mode. In the test mode, the data reliability was indeed guaranteed.
2. bp8_facts: Set innodb_buffer_pool_instances = 8 and metadata_locks_hash_instances = 256. The default configuration is innodb_buffer_pool_instances = 1.
3. nobinlog: indicates the mode in which binlog is disabled.

My opinion: the official version is getting better and better. In addition to maintaining the advantage in ease of use, the Percona branch is getting weaker and weaker in terms of performance improvement. If this is the case, you can switch to the official version.

Edit recommendations]

1. Wikipedia escaped from MySQL and supported open source databases
2. MariaDB has become a turning point in the fate of MySQL?
3. Rollback in case of a DELETE misoperation in MySQL
4. One-step migration from MySQL to Redis
5. Two ways to solve the problem of master-slave synchronization in mysql