Amazon began experiencing an outage on their Elastic Block Store (EBS) service in the early hours of April 21, 2011. This event triggered outages of other Amazon services, including the Elastic Compute Cloud (EC2) service. Several well-known websites and companies like Reddit, Palo Alto Software, Wildfire, Hootsuite, and Heroku, were victims out of the outage.
“This event reinforces the need for proper, thorough business continuity (BC) and disaster recovery (DR) planning regardless of where your critical servers live,” said Derek Silva, Research Analyst, Info-Tech Research Group. “Several of the high profile outages were for Amazon clients who located all of their cloud infrastructure in one Amazon availability zone.”
Amazon availability zones are essentially isolated physical and logical resource pools within the larger Amazon cloud. Effective DR and BC planning includes redundancy across multiple locations.
“This means mirrored servers at one of Amazon's other data centers, or perhaps even on another compute cloud. It's entirely possible to do that, and it's exactly the type of thing that needs to be done in order to really ensure redundancy and high availability." said Silva.
Cloud services are inexpensive and generally easy to deploy, but putting services and servers in a cloud doesn't preclude enterprises from building solutions for when an outage does inevitably happen.