Well, not quite, but there’s definitely a stormy front hitting the cloudosphere with the recent reports (originally here) of Amazon’s pending EC2 instances’ reboots. Although the exact details are yet to be known, it looks like a good number of Amazon’s server instances are scheduled for compulsory reboot due to maintenance reasons within the coming weeks. This is, understandably, unpleasant news for anyone using AWS.
We are using tens of Amazon’s servers to power (and test and develop) our Database-as-a-service solution. Judging from our inventory, here’s what we do know:
- More than 80% of our servers are scheduled for reboot
- The maintenance-bound instances are both in the us-east as well as the us-west regions
- It looks like only the newer instances (those started in last couple of weeks) are exempt
Like earlier events, this is another reminder of the fact that availability in the cloud shouldn’t be taken for granted and if there’s one thing to expect it’s the unexpected. Unlike previous outages, this time it’s a planned maintenance event and Amazon is doing a great job (at least in our case) in keeping us informed, providing documentation and assisting us to cope with the maintenance window.
The good news are that since these are instance reboots (and not system reboots), we’ll actually be able to control the schedule of reboots ourselves and keep the risk of any interruptions to our service to the absolute minimum. Leveraging on our DBaaS’s built-in availability mechanisms, we don’t expect our users to experience any impacts as a results of this reboot cycle. We’re planning to do this next week and we’ll be sure to keep you updated on our progress.
So the countdown has begun: little more than a week to go!
And yes: we shall overcome! :)