PolarDB for MySQL:PolarProxy Enterprise Edition

Architecture

A PolarDB cluster of Cluster Edition and X-Engine Edition consists of a primary node and up to 15 read-only nodes. By default, PolarDB provides two types of endpoints: the primary endpoint and cluster endpoints. The cluster endpoints are provided by PolarDB PolarProxy. The cluster endpoint can be in read and write (automatic read/write splitting) or read-only mode.

PolarProxy Enterprise Editions

Billing

Limits

Only clusters of PolarDB Cluster Edition and X-Engine Edition support cluster endpoints and PolarProxy. Clusters of PolarDB Single Node Edition use a single-node architecture, and do not support PolarProxy and cluster endpoints.

Precautions

• PolarProxy does not support compression protocols for the default and custom cluster endpoints.
• If you do not enable the transaction splitting feature after cluster endpoints are used, all the requests in a transaction are forwarded to the primary node.
• When you execute the SHOW PROCESSLIST statement after cluster endpoints are used, the system returns the results of all the nodes.
• If you execute multi-statement or call stored procedures, all subsequent requests for the current connection are forwarded to the primary node. For more information, see Multi-Statement. To use the read/write splitting feature, you must close the current connection and reconnect to the cluster.
• The maximum number of connections to a PolarDB cluster endpoint depends on the specifications of compute nodes in backend databases. For a cluster endpoint that supports read/write splitting, an application establishes a connection to each compute node in the backend database. Therefore, the maximum number of connections that an application can use is the maximum number of connections of a single compute node. For a cluster endpoint that is in read-only mode, an application connection only establishes a connection to one compute node in the backend database. The maximum number of connections that an application can use is the sum of the maximum numbers of connections for all read-only nodes. For a cluster endpoint that is in read-only mode, you can use the Transaction-level connection pools feature to increase the maximum number of connections that an application can use.
• If a session that supports read/write splitting is created after you add or restart a read-only node, read requests are forwarded to the read-only node. If a session that supports read/write splitting is created before you add or restart a read-only node, read requests are not forwarded to the read-only node. To forward these read requests to the read-only node, you must close the connection and reconnect to the cluster. For example, you can restart your application to establish a new connection.
• Do not modify environment variables when you call stored procedures or execute multi-statement, such as set names utf8mb4;select * from t1;. Otherwise, the data read from the primary node and read-only nodes is inconsistent.
• To prevent the application connection pool from being hung due to packet loss caused by network issues or other unexpected scenarios when you use the application connection pool to connect to a URL, we recommend that you add the connectTimeout and socketTimeout parameters to the URL. For more information about the parameters, see Parameters. Example:

  java url: jdbc:mysql://ip:port/db?connectTimeout=60000&socketTimeout=300000;

Related API operations