At the recent London Cloud Camp I started thinking about the less obvious downsides to Cloud Computing and a move to a more centralised computing platform. The benefits to this approach are obvious: economies of scale, reduced capital outlay and lower barrier to entry.

What about the downsides? One of the arguments from IT departments is the explicit risk measurement that can be achieved through the outsourcing of infrastructure to the cloud. Instead of the systems team having to write risk management plans and defining how likely downtime is and how much it would cost the cloud provider offers an explicit guarantee in the form of an SLA. The management of that risk is now in the hands of the cloud provider and the company just needs to pay them their fee. Surely this is a good thing? Well, yes, to a certain extent: relying on someone else to manage your risk isn’t always the best approach. Consider the current economic crisis caused by badly managed risk. Banks were caught out because they thought there was less risk involved in lending than the reality.

The other thing that exacerbated the credit crunch was the network effect. Banks in different countries had become so intertwined, everyone had lent and borrowed money from everyone else. When banks asked for their money back it caused a chain reaction of more banks asking for their money back from each other. A cloud computing world is more interconnected and reliant on one or two providers (theoretically, these providers should be more resilient because of multiple data centres) so if something goes wrong then the damage caused is more likely to be greater than a single company’s infrastructure being compromised.

These two effects (the network effect and the bundling up of risk) are important for the future of cloud computing. Cloud providers must offer transparency and detailed SLAs that mitigate the effect of unforseen events. Companies must also realise that going on the Cloud is by no means a Silver Bullet and must take into account all the risks involved (and realise that extreme and unforseen events are more likely to cause problems than when using their own infrastructure).