How to Understand Galera Replication

Replication and High Availability Strategies
24 April 4:30pm - 5:20pm @ Ballroom B

Galera was conceived from the ground up to be the simplest replication solution possible. And it really is very simple. But it is also very different from tradional replication approaches, and thus may be non-trivial to grasp at first sight.

If you looked at Galera replication for MySQL (Percona XtraDB Cluster, MariaDB Galera Cluster), even tried it, and asked one of the following questions, it means you are thinking in terms of old-school MySQL replication or some other replication technology other than Galera:

* What happens when the primary node dies?
* Does Galera use 2 phase commit?
* What happens when a transaction is committed on the master but fails on the slave?
* When all nodes have certified the transaction, do they then "approve" the commit back to the master?
* How can I setup read-only slaves?
* How do I promote another node to master? How does failover happen?
* If a node is disconnected from the cluster, how can I prevent my application from writing to it to avoid inconcistent slave issues?

If you are trying to ask the above questions about Galera, then you should attend this talk - it will help you to make sense out of it all.


Alexey Yurchenko
Solutions Architect, Codership Oy
Alexey Yurchenko is a founder and a self-appointed Solutions Architect at Codership Oy, developers of Galera replication engine.
Seppo Jaakola
CEO, Codership
Database replication specialist, with long track record of developing various replication and clustering solutions for MySQL and PostgreSQL databases. Currently holding CEO position at Codership, working for Galera open source project with the target of developing a universally applicable synchronous multi-master replication solution.