With IBM, Amazon and Microsoft fighting for our cloud computing attention, we might forget to think about useful scenarios for all this fancy airy stuff. We all know the classical e-commerce example with its season-bound resource spikes and the occasional raw processing/render power we need, but this week I stumbled upon something new: cloud bursting.
Before I get into that, let’s first take a little step back. A recent blog post by James Urquhart (now helping Cisco in cloud computing strategy) lets us wonder whether a cloud computing strategy is always cheaper than having your own data center. As you might have guessed: no. The main idea is that applications that are always running on 100 % are often hosted cheaper in your own data center than paying for the elasticity of a cloud (which you actually don’t use).
So, applications that do need the elasticity (e.g. the e-commerce and render farm we talked about earlier) can benefit financially from the cloud. One other strategy that I didn’t think of earlier (stupid me) was pointed out on Scobleizer’s post of his visit to the Cisco data center: cloud bursting. Simply said: run everything in your own data center and once you detect heavy usage, offload it to the cloud! Example he gives is 12seconds.tv that offloads a movie to Amazon’s S3 file storage when they detect high view rates.
Can we think of any useful applications in the enterprise space? Oh yes: update / file servers (it only gets busy when you release a new product or service pack), payroll processing (only once a month, but do you want this in the cloud?), … Any other suggestions?