Table of Contents [+/-]
The MySQL Query Analyzer enables you to monitor SQL statements executed on a MySQL server and see details about each query, number of executions and execution times. Similar queries with different literal values are combined for reporting purposes.
Query Analyzer collects information about SQL statements that a MySQL client application sends to the MySQL server. There are different methods that the Query Analyzer can receive this information, which are:
Using the Performance Schema statement digests with MySQL Server 5.6.14 and above, data can be gathered directly from MySQL Server without additional configuration, using a MySQL Enterprise Monitor Agent.
The client application can route its database requests through the Proxy and Aggregator. The Proxy routes the client's query to both the MySQL instance and the Aggregator. The Aggregator normalizes the queries and transmits them to the Service Manager.
Install a MySQL Enterprise Monitor Plugin for a Connector that sends the information directly to MySQL Enterprise Service Manager.
Once your MySQL client application is configured to communicate via the MySQL Enterprise Monitor Agent, queries are monitored and the normalized queries are sent to the MySQL Enterprise Monitor Agent.
For the different ways to enable Query Analysis, see Section 32.1, “Providing Query Analyzer Data”. For the user interface of the Query Analyzer, see Chapter 32, Using the Query Analyzer and Section 32.3, “Query Analyzer User Interface”.
Once the data is collected, you view and monitor the queries, check the execution statistics, and filter and drill down on the information. By comparing the queries to the server graphs, you can correlate query execution with server status. For more information on viewing, filtering and reporting on the Query Analyzer data, see Section 32.3, “Query Analyzer User Interface”.
When MySQL Enterprise Monitor is not accessible from a
query analyzer plugin, the application performance is not
impacted. Over time, the plugin determines that a backlog of
reportable data exists, and fall back to consolidating it over
longer ranges of time. But if more than 1,000 canonical queries
are being used by the application (an unlikely scenario), data is