Amazon AWS OutageOctober 29, 2012
Author: Rich Turner
Amazon’s North Virginia data center experienced a serious service disruption recently taking down many popular websites, including Pinterest, Reddit, Foursquare, Minecraft , and seriously affecting others, like GitHub, Flipboard, Instagram and Airbnb – to name but a few.
This outage, as well as many other high profile cloud outages, amplifies the resilience challenges of hosted services, and leaves me to ask not if on premise is better than cloud, but rather how much control am I willing to relinquish by moving to cloud services? And, is it worth the risk when an outage does occur?
Let me explain. If a service that I am responsible for goes down, I want to know 1) people are working to rectify the issue, 2) when services will be available, and 3) what is the impact of the outage. Ultimately, I want assurance that the matter is being managed as if I were in full control. Giving up control, and relying on a few snippets on a ‘current status’ webpage scares the hell out of me. Having to tell the rest of the affected business that I don’t know what is being done, when it will be done or what the impact will be is not an acceptable option.
After all, what confidence do I really have that the last status update is accurate? How do I know that the service provider is not currently huddled in a conference room working out the best strategy to inform the world that all data is lost and unrecoverable? This is an extreme example, but it’s happened in the past.
ArchiveOne Enterprise has a number of cloud storage options, because we understand the many benefits of cloud for our customers. However, we balance cloud benefits with service control with our hybrid offering. Store your archive data in the cloud if the risk, when weighed against the benefits, is acceptable. Or, store everything on premise, if you don’t. Alternately, maintain an on-premise copy and a backup copy in the cloud. Of course, should you change your mind about cloud in the future, you can migrate it all back to on premise with 2 clicks in the Administration Console. In any scenario, maintain control with on-premise software – the best of both worlds.
Let me finish with this very well made point from Ars Technica :
The takeaway for anyone considering cloud hosting for your website or application is that there is no magic pill, and just because something is “in the cloud” doesn’t mean it can’t come crashing down to earth when there are problems.