Updated on 2024-04-03 GMT+08:00

Concurrency Control

Scenarios

You can create rules to control concurrent execution of SQL statements by specifying SQL type, keywords, and maximum concurrency. To maintain better performance at high concurrency, SQL statements that meet the specified SQL type and keyword and exceed the maximum concurrency will not be executed.

High SQL concurrency can be caused by the following factors:

  • A sharp increase in requests: Concurrent SQL statements of a certain type surge due to cache penetration and abnormal calls.
  • Stacked slow queries: If a large number of SQL statements without indexes are called, many slow SQL statements will be generated, affecting services.

Supported Versions

Concurrency Control is available to the RDS for MySQL versions listed in Table 1.

Table 1 Supported versions

Major Version

Minor Version (Primary Instance)

Minor Version (Read Replica)

Setting Rules for Read Replicas Separately

5.6

≥ 5.6.50.3

≥ 5.6.51.6

Not supported

5.7

≥ 5.7.31.4

≥ 5.7.37.1

≥ 5.7.38.221000

8.0

≥ 8.0.25.1

≥ 8.0.25.1

Not supported

In some versions, concurrency control rules are not applied to requests sent by user root. For details, see Table 2.

If your instance running 5.7.43.231000 or later or 8.0.28.231000 or later, concurrency control rules will be applied to user root. To disable concurrency control for user root, contact customer service.

Table 2 Versions in which requests from root are not limited by concurrency control rules

Major Version

Minor Version (Primary Instance)

5.6

≥ 5.6.51.4

5.7

5.7.33.1 ≤ Version < 5.7.43.231000

8.0

8.0.25.1 ≤ Version < 8.0.28.231000

To achieve better performance at high concurrency, you are advised to upgrade your DB instance to the latest minor version. For details about how to upgrade a minor version, see Upgrading a Minor Version.

Constraints

  • A maximum of 100 concurrency control rules can be configured.
  • Only SELECT, UPDATE, and DELETE statements are supported for concurrency control. INSERT statements are not supported.
  • If a SQL statement matches multiple concurrency control rules, only the most recently added rule is applied.
  • SQL statements that have been executed before a concurrency control rule is added are not counted.
  • If the replication delay is too long, adding or deleting a concurrency control rule for a read replica does not take effect immediately.
  • Concurrency control rules are not applied to system tables.
  • Concurrency control rules are not applied to SQL statements not used for data query, such as select sleep(***);.
  • Concurrency control rules are not applied to stored procedures, triggers, or functions.
  • You can run the following SQL statement through DAS to view the execution of concurrency control rules: select * from information_schema.rds_sql_filter_info;
  • Too many concurrency control rules affect the database performance. Delete unnecessary rules after using them.

Procedure

  1. Log in to the management console.
  2. Click in the upper left corner and select a region and a project.
  3. Click in the upper left corner of the page and choose Databases > Relational Database Service.
  4. On the Instances page, click the target DB instance name.
  5. In the navigation pane, choose DBA Assistant > Historical Diagnosis.
  6. Choose SQL Explorer > Concurrency Control.
  7. Toggle on the concurrency control switch .

    Concurrency control rules take effect only after concurrency control is enabled.

  8. Click Add Rule. Configure the parameters listed in Table 3.

    Figure 1 Adding a rule
    Table 3 Parameter description

    Parameter

    Description

    SQL Type

    There are three options: SELECT, UPDATE, and DELETE.

    Keyword

    A maximum of 128 keywords (case-insensitive) are supported. You can specify keywords in either of the following ways:

    • Enter keywords: Take select~a as an example. select and a are two keywords contained in a concurrency control rule. The keywords are separated by a tilde (~). In this example, the rule restricts the execution of only the SQL statements containing keywords select and a.
    • Generate keywords from a SQL statement: You can enter a SQL statement and then click Generate Keyword. The generated keywords are for reference only. Exercise caution when using them.

    SQL statements match the keywords from first to last. For example, if one rule contains the keyword a~and~b, the statement *** a>1 and b>2 can match the keyword, but *** b>2 and a>1 cannot.

    Empty characters before and after each keyword will be ignored, for example, spaces, '\n', '\r', and '\t'.

    Max. Concurrency

    If the number of concurrent SQL statements matching the keyword exceeds this limit, the SQL statements will not be executed. The value ranges from 0 to 1,000,000,000.

    Kill existing sessions that match this rule

    If this option is selected, all sessions generated by users subject to this concurrency control rule will be killed.

    For details about the versions where user root is not subject to concurrency control rules, see Table 2.

  9. Confirm the settings and click OK.

Follow-up Operations

To delete a concurrency control rule, locate it in the rule list and click Delete in the Operation column. In the displayed dialog box, click OK.

Figure 2 Deleting a rule