AWS Outage. Why our clients where not impacted
Back in late April 2011, many companies where directly impacted by the outage of one of Amazon’s AWS regions. We noticed quite a few unhappy people blogging on the web. The problem here was not really that AWS had a region outage but rather, those impacted had used AWS as merely a hosting service. so in essence, you could have hosted your web applications on any hosting provider and had the same problem.
AWS is a cloud service, and offers various regions that are isolated.
OK, So what is the difference? Quite simply, AWS provides a means for easy distribution across multiple regions. The systems we support are generally run from one region but all backups and disaster recovery measures are within one or more other region.
When we received our first set of alerts, we knew that the impacted AWS region had issues and began our DR process for our customers in that region.
DR is really easy once you have put the effort into automating as much as possible and AWS gives you the tools to do this, well some of the tools anyway.
For example, one of our Mobility customers was hosted within this region. Once we received the alerts we executed DR. The replicated instances where activated, the latest data backup restored from S3 and the Elastic IP Addresses redirected. This took less that 1 minute in total and our customers did not notice a single thing, until we advised them.
So the lesson in all of this is, if you plan to use AWS just as a hosting provider, then be prepared to suffer the same impacts any hosting provider is at risk of. AWS provides the ability to host in various parts of the world, keep the backup instances shutdown, reducing your costs, and have a pre planned set of scripts ready to go. Hence, no loss of income, no unhappy consumers and the directors and managers can all rest easy.