In an interview with Computing.co.uk the LOCOG CIO Gerry Pennell revealed that “the infrastructure in the cloud is not sufficiently mature enough to support the kinds of things we're doing in the Olympics” and had rejected a cloud based solution. It’s difficult to blame Pennell for declining cloud in this way, core infrastructure systems around logistics, communications and broadcast would need to be incredibly robust to satisfy all the requirements of such a big operation.
Adoption of any technology takes time. London 2012 will arguably be the first ‘social’ Olympics; though many called Beijing exactly that in 2008, the proliferation of social has caused commentators to use the term again to describe this year’s games. So we can’t necessarily call this the first ‘Cloud Olympics’ which is a shame, as cloud theoretically meets the needs of a big sporting event such as this.
The infrastructure is a time-limited one, operational for a few months prior to the event and not long after. Only temporary hosting and compute power is required and so cost savings could be found there. Scalability for accessibility is important. Live events such as the Olympics will experience ‘peaking’ where demand is extremely high in short bursts. Having access to extra scalability quickly could be a key cloud advantage. Security would be crucial, but could not be managed via legacy, rigid security models. Dynamic security policies aware of the changing infrastructure and capable of adapting would be required.
It’s clear that the most flexible ‘suits all’ approach would be to adopt a hybrid model, but outages, such as those experienced in the public cloud recently, would be unacceptable. It will be interesting to see how the adoption changes for Rio 2016 and beyond.