Architecting OpenStack for High Availability

Reliability and availability are challenges for any production IT deployment, but when it comes to cloud architecture, high availability (HA) is a key consideration. Enterprises that are deploying OpenStack in real-world production environments often have to learn how to build in HA as they go along. Because RightScale has been helping organizations manage OpenStack clouds since the project's first days, we're in consideration to be chosen to present on best practices and in-the-trenches stories for a session at OpenStack Summit in Hong Kong in November, and this month the community gets to vote on who speaks. Here's a preview of what I hope to talk about in a session entitled Stacking Up with OpenStack: Building for HA.

The first consideration is to start with a cloud deployment architecture that handles hardware, software, and network failure. That means having no single point of hardware failure, and designing stateless applications that won't crash when a single server goes down. You can then optimize and automate your applications for high availability within a single OpenStack cloud or across regions.

You also have to plan what to do when a resource fails. For each application, you should determine a recovery time objective (RTO) — a time period in which service must be restored to meet business continuity planning objectives — and a recovery point objective (RPO) — an acceptable data loss as a result of recovering from a disaster or catastrophic event. One has to keep the cost vs. RTO/RPO in mind. Depending on your specific requirements and cost tolerance, we'll suggest various HA/DR strategies to employ, including multi-region OpenStack clouds and hybrid clouds.

One key requirement is the need to build in automation, so that once you've identified a failure, you can recover from it quickly. With OpenStack, you can get some of the same type of automation that's available to public cloud users on services such as Windows Azure, Google Cloud Platform, and Amazon Web Services (AWS).

If you'd like to hear me present  at OpenStack Summit, please vote for Stacking Up with OpenStack: Building for HA. I'm looking forward to seeing you in Hong Kong.