MySQL Fabric is a system for managing a farm of MySQL servers (and other components). Fabric provides an extensible and easy to use system for managing a MySQL deployment for sharding and high-availability.
Connector/Python 1.2.0 and up supports Fabric and provides these capabilities:
Requesting a connection to a MySQL server managed by Fabric is as transparent as possible to users already familiar with Connector/Python.
Connector/Python is able to get a MySQL server connection given a high-availability group and a mode specifying whether the connection is read-only or also permits updates (read-write).
Connector/Python supports sharding and is able to find the correct MySQL
server for a given table or tables and key based on scope
(local or global) and mode (read-only or read-write).
mechanisms are supported transparently to the user.
Among secondary MySQL servers in the same group, load balancing of read-only operations occurs based on server weight.
Faulty MySQL servers are reported to Fabric, and failover is supported when failure occurs for a server in a group.
Connector/Python caches information coming from Fabric to speed up operations. Failures connecting to a MySQL server reset this cache. The time to live for cached information can be set by Fabric, or a default is used otherwise.
Fabric support applies to versions of Python supported by Connector/Python itself (see Chapter 3, Connector/Python Versions). In particular, you can use Connector/Python with Python 3.1 and later to establish Fabric connections, even though Fabric does not support Python 3.
For more information, see MySQL Fabric.