AnalyticDB for MySQL uses different backend storage systems to store collected monitoring data based on data types.

Time series data storage

Time series databases have optimized the writing, aggregation, and retrieval of time series data. They are used to store the time series data collected by AnalyticDB for MySQL. Each AnalyticDB for MySQL cluster has an independent time series database that stores only the time series data of the current cluster.

Log data storage

Log data includes audit logs and query details logs. After logs are disassembled, combined, and aggregated, the logs are stored in online analytical processing (OLAP) databases for subsequent real-time multidimensional analysis. The audit logs of a cluster are stored in a separate database table. The detailed information of audit logs includes the user, database, response time, SQL statement, and type. After the detailed information of audit logs is aggregated from multiple dimensions, profile table information of the cluster is generated and stored in a profile image table. Query details logs are divided into statistics tables of the following levels: query, subtask, operator, and plan node. They can be aggregated in different manners to solve different system issues.