Memcached replication for data redundancy

A lot of high performance web sites use memcached to speed up caching and serve content much faster to the web visitors. By default memcached doesn’t offer any data redundancy and no replication feature – memcached uses server RAM to store all data and if the server hardware fails or it is rebooted all data that’s stored in the memcached daemon will be lost. In order to increase memcached data redundancy and offer high availability you should check out RepCached (Web site: http://repcached.lab.klab.org/ ) – it offers multi-master and asynchronous replication for 1.2.x memcached versions.

You can use Layer-4 load balancer like Linux Virtual Server to balance load and offer high-availability between multiple memcached (with repcached patches) instances on different physical servers. Then you can add a script that removes failed memcached instances from LVS balanced pool and you will have a high-availability, load balanced memcached cluster.


Leave a Reply