Cloudbursting allows failover into the cloud

Cloudbursting allows failover into the cloud

This is a link to a podcast who's primary focus is to generate interest in F5 and how they can serve cloud computing but I like the analogy Jason Meserve uses to describe cloudbursting. He compares it to overdraft protection on your bank account. When data center resources are exhausted for a particular application/system, you can push it to the cloud which essentially gives you on-demand provisioning. While this is somewhat of a sidebar for cloud computing, I think it does well to make you think about ways to commodotize cloud computing.

To put it into perspective, I'll use a recent example with one of our customers. Anyone who follows Hosted Solutions knows that we recently signed Belk department stores as a customer (belk.com). Just before the big Thanksgiving shopping weekend, Belk had some concerns about whether they would be able to handle the shopping load to their site. Obviously when you put a lot into your e-Commerce environment, you want it to be able to handle heavy traffic when needed. Otherwise, you lose money. Members of my team spent a lot of late nights helping to make sure Belk could handle their needs the day after Thanksgiving. We worked with third party companies that essentially model customer load so that you can see how your systems react. At the end of the day, we ended up adding some additional resources in the Belk environment so that it could scale to an extreme amount of users in the event they got slammed with shoppers.

From a cloud computing standpoint, think about how this could have been different. If we had a cloud where we could instantly call upon resources when needed and the management interfaces necessary to harvest those resources quicly and efficiently, we could have allowed the existing Belk infrastructure to run and only provisioned more resources if we needed to. Best case scenario, Belk would have had a portal to view their customer traffic load, set threshholds where if traffic reached certain levels they would have been alerted and they would have had the ability to instantly order additional cloud resources to help with their volume. That order would have kicked off automatic processes that replicated their data/site and within a few minutes (or sooner) they would have the capacity to deal with their spike in customer traffic volumes. All of this of course could be automated as well.

Working for a company that provides data center services, I personally think this could be a way to get "cloud's foot in the door" so to speak. If you can offer companies the ability to provision additional resources on a dime, then you don't waste money or engineering cycles with load testing and trying to estimate how much volume will increase. Who cares at that point? If it goes up, they'll be able to respond instantly and meet their business challenges.