Preparing for the Big Oops: How to Build Disaster Recovery Sites for MySQL

At Continuent we have deployed a number of MySQL DR sites and have been gathering up our experience into a cookbook procedure. In this talk we will share our experience and give you a set of steps to help you get your own DR site up and running in no time using off-the-shelf technologies like MySQL replication and/or Tungsten replication.

Topics covered:

• Standard disaster recovery architectures * Site capacity questions * Physical sites vs. cloud * Writable vs. Read-only DR sites * Database replication vs. DRBD vs. fancy hardware
• Steps to set up a main + DR site configuration * Replication configuration, backups, and day-to-day management
• Failing over * Declaring a site outage * Making the DR site writable * Fencing the old site * Failing over applications
• Failing back to the main site * Recovering lost data from the main site * Resynchronizing hosts across sites * Scheduling and managing failback to the main site
• Making it work * How to test DR up-front * How to test DR on an on-going basis
• Summary and Questions

Along the way we’ll try to remember every gotcha we have seen so you don’t have to hit them yourself. At the end of this talk you’ll have a solid plan for setting up your own DR site quickly and efficiently.