The big storms yesterday in the eastern part of the country took down part of Amazon EC2 cloud services in Virginia due to the power outages. The Amazon EC2 outage of course had a cascading effect on customers including some high profile names like Netflix, Pinterest and Instagram. Fortunately, Amazon EC2 recovered quickly and everyone was back up in about 4½ hours. Even though the power outage didn’t last for very long, this wasted valuable viewing pleasure for audiences across the country. These people may decide to buy a TV box in the hopes that their programs don’t get interrupted again. What’s even better, is that they will have access to more streaming services so that they are able to watch something else if they have had to wait for the power outage to be resolved.
Since I am sure a few colleges and universities were also affected by the Amazon EC2 outage, I imagine the higher ed forums will light up with Cloud FUD (Fear, Uncertainty & Doubt).
Even if the impacts to colleges were incidental. Friday’s outage will still be used in arguments against further use of Amazon Web Services, cloud services in general or other forms of outsourcing to the cloud. Especially because Amazon EC2 had a bit a of a hick-up just two weeks ago.
Don’t misunderstand this. It is important to continue the discussions on improving the reliability of cloud services and building redundancy into an enterprise cloud architecture. It is also to important that we devote time to discussing risk mitigation strategies for cloud services.
Just don’t light your hair on fire or start channeling Chicken Little.
But let us also not lose site of the importance of also continuing the discussions on the campus power outages that also occurred. Let’s be sure we devote a commensurate, and proportionate, amount of energy and thought to the cascading events from losing commercial power and disruptions to on-premise technology based services remembering of course we have seen this drill before and should have it mastered.
There are of course data center outages from failed transfers to back-up then emergency power and perhaps failures from prolonged operations of emergency power systems or overload conditions from poor capacity planning.
But there are also failures outside the data centers in emergency notification systems and processes reliant on third party providers and carriers. And so on down the line for wireless access, IDF’s and departmental computer rooms even remote applications.
For some campuses still suffering from a loss of commercial power like many in Ohio, their test may just be beginning as they face a protracted outage. This presents an unfortunate opportunity to see if anyone applied the lessons learned from the devastation in the I-95 corridor from Tropical Storm Lee last September or not.
If there is a silver lining it is that unlike Pinterest or Netflix, campuses have the luxury of shutting down and cancelling events over the weekend or until commercial power is restored.