Fast Failover using MySQL and Zookeeper
Replication and High Availability Strategies
4 April 3:00PM - 3:50PM @ Ballroom E
50 minutes conference
Fast failover is not just an option for building reliable systems in the cloud - it is a requirement. Unfortunately, it seems like every storage solution out there (whether it's SQL or NoSQL) has its own proprietary monitoring and failover mechanism. In this presentation, we explore Zookeeper as a mechanism for fast, application-driven failover for both MySQL and Redis in the cloud. By using Zookeeper and client-level integration, we can avoid the "magic" of network-level failover (such as elastic IPs) as well as the latency and complexity of proxy-level solutions. We demonstrate that Zookeeper is a practical cross-platform strategy from any language, with examples of failover in Ruby, Java, and Node.js applications.
Sunny Gleason is founder and Distributed Systems Engineer at SunnyCloud, a company providing high-performance application development and hosting and high-availability distributed storage solutions using Amazon Web Services, Rackspace, and Google Compute Cloud. Previously, he worked on cloud platform system design and implementation at Amazon.com and Ning. LinkedIn: http://www.linkedin.com/in/sunnygleason/ Twitter: http://twitter.com/sunnygleason GitHub: http://github.com/sunnygleason