30.4.4.25 The statement_performance

30.4.4.25 The statement_performance_analyzer() Procedure

Creates a report of the statements running on the server. The views are calculated based on the overall and/or delta activity.

This procedure disables binary logging during its execution by manipulating the session value of the sql_log_bin system variable. That is a restricted operation, so the procedure requires privileges sufficient to set restricted session variables. See Section 7.1.9.1, “System Variable Privileges”.

Parameters

in_action ENUM('snapshot', 'overall', 'delta', 'create_tmp', 'create_table', 'save', 'cleanup'): The action to take. These values are permitted:
- snapshot: Store a snapshot. The default is to make a snapshot of the current content of the Performance Schema events_statements_summary_by_digest table. By setting in_table, this can be overwritten to copy the content of the specified table. The snapshot is stored in the sys schema tmp_digests temporary table.
- overall: Generate an analysis based on the content of the table specified by in_table. For the overall analysis, in_table can be NOW() to use a fresh snapshot. This overwrites an existing snapshot. Use NULL for in_table to use the existing snapshot. If in_table is NULL and no snapshot exists, a new snapshot is created. The in_views parameter and the statement_performance_analyzer.limit configuration option affect the operation of this procedure.
- delta: Generate a delta analysis. The delta is calculated between the reference table specified by in_table and the snapshot, which must exist. This action uses the sys schema tmp_digests_delta temporary table. The in_views parameter and the statement_performance_analyzer.limit configuration option affect the operation of this procedure.
- create_table: Create a regular table suitable for storing the snapshot for later use (for example, for calculating deltas).
- create_tmp: Create a temporary table suitable for storing the snapshot for later use (for example, for calculating deltas).
- save: Save the snapshot in the table specified by in_table. The table must exist and have the correct structure. If no snapshot exists, a new snapshot is created.
- cleanup: Remove the temporary tables used for the snapshot and delta.
in_table VARCHAR(129): The table parameter used for some of the actions specified by the in_action parameter. Use the format db_name.tbl_name or tbl_name without using any backtick (`) identifier-quoting characters. Periods (.) are not supported in database and table names.
The meaning of the in_table value for each in_action value is detailed in the individual in_action value descriptions.
in_views SET ('with_runtimes_in_95th_percentile', 'analysis', 'with_errors_or_warnings', 'with_full_table_scans', 'with_sorting', 'with_temp_tables', 'custom'): Which views to include. This parameter is a SET value, so it can contain multiple view names, separated by commas. The default is to include all views except custom. The following values are permitted:
- with_runtimes_in_95th_percentile: Use the statements_with_runtimes_in_95th_percentile view.
- analysis: Use the statement_analysis view.
- with_errors_or_warnings: Use the statements_with_errors_or_warnings view.
- with_full_table_scans: Use the statements_with_full_table_scans view.
- with_sorting: Use the statements_with_sorting view.
- with_temp_tables: Use the statements_with_temp_tables view.
- custom: Use a custom view. This view must be specified using the statement_performance_analyzer.view configuration option to name a query or an existing view.

Configuration Options

statement_performance_analyzer() operation can be modified using the following configuration options or their corresponding user-defined variables (see Section 30.4.2.1, “The sys_config Table”):

debug, @sys.debug
If this option is ON, produce debugging output. The default is OFF.
statement_performance_analyzer.limit, @sys.statement_performance_analyzer.limit
The maximum number of rows to return for views that have no built-in limit. The default is 100.
statement_performance_analyzer.view, @sys.statement_performance_analyzer.view
The custom query or view to be used. If the option value contains a space, it is interpreted as a query. Otherwise, it must be the name of an existing view that queries the Performance Schema events_statements_summary_by_digest table. There cannot be any LIMIT clause in the query or view definition if the statement_performance_analyzer.limit configuration option is greater than 0. If specifying a view, use the same format as for the in_table parameter. The default is NULL (no custom view defined).

Example

To create a report with the queries in the 95th percentile since the last truncation of events_statements_summary_by_digest and with a one-minute delta period:

Create a temporary table to store the initial snapshot.
Create the initial snapshot.
Save the initial snapshot in the temporary table.
Wait one minute.
Create a new snapshot.
Perform analysis based on the new snapshot.
Perform analysis based on the delta between the initial and new snapshots.

mysql> CALL sys.statement_performance_analyzer('create_tmp', 'mydb.tmp_digests_ini', NULL);
Query OK, 0 rows affected (0.08 sec)

mysql> CALL sys.statement_performance_analyzer('snapshot', NULL, NULL);
Query OK, 0 rows affected (0.02 sec)

mysql> CALL sys.statement_performance_analyzer('save', 'mydb.tmp_digests_ini', NULL);
Query OK, 0 rows affected (0.00 sec)

mysql> DO SLEEP(60);
Query OK, 0 rows affected (1 min 0.00 sec)

mysql> CALL sys.statement_performance_analyzer('snapshot', NULL, NULL);
Query OK, 0 rows affected (0.02 sec)

mysql> CALL sys.statement_performance_analyzer('overall', NULL, 'with_runtimes_in_95th_percentile');
+-----------------------------------------+
| Next Output                             |
+-----------------------------------------+
| Queries with Runtime in 95th Percentile |
+-----------------------------------------+
1 row in set (0.05 sec)

...

mysql> CALL sys.statement_performance_analyzer('delta', 'mydb.tmp_digests_ini', 'with_runtimes_in_95th_percentile');
+-----------------------------------------+
| Next Output                             |
+-----------------------------------------+
| Queries with Runtime in 95th Percentile |
+-----------------------------------------+
1 row in set (0.03 sec)

...

Create an overall report of the 95th percentile queries and the top 10 queries with full table scans:

mysql> CALL sys.statement_performance_analyzer('snapshot', NULL, NULL);
Query OK, 0 rows affected (0.01 sec)

mysql> SET @sys.statement_performance_analyzer.limit = 10;
Query OK, 0 rows affected (0.00 sec)

mysql> CALL sys.statement_performance_analyzer('overall', NULL, 'with_runtimes_in_95th_percentile,with_full_table_scans');
+-----------------------------------------+
| Next Output                             |
+-----------------------------------------+
| Queries with Runtime in 95th Percentile |
+-----------------------------------------+
1 row in set (0.01 sec)

...

+-------------------------------------+
| Next Output                         |
+-------------------------------------+
| Top 10 Queries with Full Table Scan |
+-------------------------------------+
1 row in set (0.09 sec)

...

Use a custom view showing the top 10 queries sorted by total execution time, refreshing the view every minute using the watch command in Linux:

mysql> CREATE OR REPLACE VIEW mydb.my_statements AS
       SELECT sys.format_statement(DIGEST_TEXT) AS query,
              SCHEMA_NAME AS db,
              COUNT_STAR AS exec_count,
              sys.format_time(SUM_TIMER_WAIT) AS total_latency,
              sys.format_time(AVG_TIMER_WAIT) AS avg_latency,
              ROUND(IFNULL(SUM_ROWS_SENT / NULLIF(COUNT_STAR, 0), 0)) AS rows_sent_avg,
              ROUND(IFNULL(SUM_ROWS_EXAMINED / NULLIF(COUNT_STAR, 0), 0)) AS rows_examined_avg,
              ROUND(IFNULL(SUM_ROWS_AFFECTED / NULLIF(COUNT_STAR, 0), 0)) AS rows_affected_avg,
              DIGEST AS digest
         FROM performance_schema.events_statements_summary_by_digest
       ORDER BY SUM_TIMER_WAIT DESC;
Query OK, 0 rows affected (0.10 sec)

mysql> CALL sys.statement_performance_analyzer('create_table', 'mydb.digests_prev', NULL);
Query OK, 0 rows affected (0.10 sec)

$> watch -n 60 "mysql sys --table -e \"
> SET @sys.statement_performance_analyzer.view = 'mydb.my_statements';
> SET @sys.statement_performance_analyzer.limit = 10;
> CALL statement_performance_analyzer('snapshot', NULL, NULL);
> CALL statement_performance_analyzer('delta', 'mydb.digests_prev', 'custom');
> CALL statement_performance_analyzer('save', 'mydb.digests_prev', NULL);
> \""

Every 60.0s: mysql sys --table -e "        ...  Mon Dec 22 10:58:51 2014

+----------------------------------+
| Next Output                      |
+----------------------------------+
| Top 10 Queries Using Custom View |
+----------------------------------+
+-------------------+-------+------------+---------------+-------------+---------------+-------------------+-------------------+----------------------------------+
| query             | db    | exec_count | total_latency | avg_latency | rows_sent_avg | rows_examined_avg | rows_affected_avg | digest                           |
+-------------------+-------+------------+---------------+-------------+---------------+-------------------+-------------------+----------------------------------+
...