One of the greatest benefit of having a common platform such as OpenStack is the ability to easily implement disaster recovery solutions across different clouds in different regions. In this presentation, we will present our experience in building the Yuruware System Mover technology, that enables an easy, configurable, self service, disaster recovery solution.
We will firstly walk through the alternative architectural scenarios, of how system replication, continuous snapshots, failover and failback, across geographic regions could be achieved, and the corresponding performance and consistency trade-offs. We then present some optimisation techniques that alleviates the challenges in engineering a total system replication - this include getting over the latency problem of moving large chunks of data across regions; automation of systems operations, and intelligently producing 'reversible operations' that can be applied in failback scenarios. We illustrate how one can easily replicate an application on the AWS EC2 cloud over to the Rackspace cloud, and dynamically failover in the event of a disaster.