Richard McManus of ReadWriteWeb posted a piece lamenting the recent spate of outages over at Amazon S3, the poster child of data in the cloud.
You know, when you combine this high profile outage with problems with sites like Twitter recently and god knows how many other sites it all begs the question as to whether the whole mashup-connect everything mentality is really the right way to go.
There’s an old adage with computers that the machine is really only as fast as the slowest component. Similarly, software is only as fast as the slowest component (either software or hardware) that it uses. When you start writing applications that are effectively an orchestration of services spread all over the globe your application will only ever be as reliable/fast/etc as the weakest link. Perhaps “card house” is a better term to use than “cloud” in that respect.
A simple mashup site that, for example, links images on S3 and Flickr to points on a GoogleMap could be brought to it’s knees and rendered completely unusable if any of those three sites goes down. Even a simple, but miscommunicated, change in the API’s of those services could cripple that application, and for that matter thousands of others out.
Perhaps our fixation with the ‘cloud’ and distribution everywhere is a little premature and silos really are the way forward. Perhaps the evolution of ‘cloud computing’ is ‘cloud aggregation’, the provision of services and technologies that make it easy to grab data from the Internet and services on it to store and work with locally, even disconnected.
