MySQL Proxy Meets: Memcache

In the past years combining MySQL with memcache focused on ‘use TRIGGERs and a memcache UDF’ or ‘do it in the app’. While one scales better and is transparent, the other is consistent.

Since MySQL 5.1 the row-based replication protocol contains all the properties to get scalability and consistency at the same time. Consistency we get by making sure the data reached a slave before we invalidate the data in memcache, scalability by keeping the query-execution free of syncronous, possibly blocking triggers.

All we need is a protocol translator which understands the MySQL replication protocol, can connect to as slave to the replication tree, filter the changes of the right tables and invalidate the data in memcache. ... and the MySQL Proxy is our foundation.