In my continuing rant about keeping away from the cloud, Amazon have just proved my point as their EC2 cloud storage system went down taking a number of web sites with it. This outage was covered on Network World:
As you can see from the text, supposed safeguards that clients paid extra for where their website or application was hosted in what are termed “multiple availability zones”, did not survive this outage. The idea being that if you application or site is in more than one of these zones if a failure occurs in one zone the other ones you are in will take over keeping you running. This however has proven to be no protection in this instance as people in multiple zones went down.
Again I would say that this is not just an Amazon issue I am sure all other cloud vendors could be hit in a similar fashion. Systems in my experience can never escape from unexpected cascading events.
If you watch the video on the page linked to above, the concept of having your applications or sites spread across multiple cloud vendors was raised and the other big issue with current cloud services raised its head that of interoperability between different vendors. If they used a standards based cloud operating system then you could theoretically have two vendors hosting your application or sites and if one goes down the other takes the strain, but I think this is a long way off. I would imagine this would be horrendously expensive as well which would put me off it even if it were possible.
My advice, stay well clear of cloud services until a mature standard emerges and unless you are a monster organisation that needs and can afford this kind of stuff possibly never use it. Keep it all under your own control so that in the event of a disaster the person who it matters most to is in charge of getting you back up and running not some 3rd party monster that sees you as a small fish in its big pond.