Scaling Databases with DBIx::Router

Making databases scale on commodity hardware requires flexibility and a bag of tricks: writing to multiple servers, splitting large tables into “shards” across servers, diverting reads to a local copy, etc. Usually this is accomplished by building an in-house layer over DBI, with hard-coded rules about dispatching queries. Then all code must be rewritten to go through this layer.

DBIx::Router provides a general solution to this problem. Because it is written on top of the new DBI::Gofer system, it can be used without changes to existing code. The dispatching of queries is specified via a set of rules that allow examination of the SQL and even the data for shard routing.