Chapter 3 Deploying MySQL Router

3.1 Bootstrapping MySQL Router
3.2 Trying out MySQL Router in a Sandbox
3.3 Basic Connection Routing
3.4 Connection Sharing and Reuse
3.5 Read/Write Splitting
- 3.5.1 Configuration
- 3.5.2 Statements
3.6 MySQL Router TLS Session Cache
3.7 MySQL Router Set Trace

Performance Recommendations

For best performance, MySQL Router is typically installed on the same host as the application that uses it. Possible reasons include:

To allow local UNIX domain socket connections to the application, instead of TCP/IP.
Note
Unix domain sockets can function with applications connecting to MySQL Router, but not for MySQL Router connecting to a MySQL Server.
To decrease network latency.
To allow MySQL Router to connect to MySQL without requiring extra accounts for the Router's host, for MySQL accounts that are created specifically for application hosts such as myapp@198.51.100.45 instead of a value like myapp@%.
Typically application servers are easiest to scale.

You can run multiple MySQL Router instances on your network, and you do not need to isolate MySQL Router to a single machine. This is because MySQL Router has no affinity for any particular server or host.

Figure 3.1 Example MySQL Router Deployment

Example deployment shows a central Group Replication and Highly Available setup with three entities pointing towards or connected to it: MySQL Shell, and two stacks that each include MySQL Router, MySQL Connector, and the Application.