Are We Really Ready to Join The Cloud?

The hottest buzz word in Information Technology today is Cloud Computing. Read any technical magazine from CIO to ComputerWorld to Information Week and the word on the street is that everyone is migrating from client/server applications running in their own data center to a cloud computing environment at an infrastructure provider. We are currently going through the same transition and technology paradigm we did back in the early 1990s when companies were migrating critical applications from large distributed systems like mainframes and AS/400s to client /server on Windows or UNIX servers.

Cloud Computing, in particular Elastic Cloud Computing, makes a tremendous amount of sense for companies that have business applications that are cyclical in demand (i.e. retailers or accounting firms). I know from experience working as a system administrator at a large retailer that on Black Friday, we really put our infrastructure and network bandwidth to test. But then there are other times in the year where the systems sit at 10-15% utilization. A big waste of green money from an environmental and financial perspective. Paying for computing resources like we do for electricity is a great concept. In terms of risk, the thinking is that data security and control of the environment are the reasons not to move critical applications to the cloud. Both risks have been addressed recently by leading vendors like Amazon, Google and others.

The reason I am asking the question if we are really ready to join the cloud is a result of the huge outage Amazon had yesterday with their EC2 platform. The outage affected the web properties for over a hundred companies. I do not know the details why there was such a big outage but my thought regarding risk in addition to security and control is availability. When working with a cloud vendor, understand the contingencies in place to make sure the cloud vendor meets service levels. It is always assumed that you should have a service and operational level agreement in place and if they are not met, then the cloud vendor should be financially responsible. One of the contingencies that should be in place are multiple electrical grids and internet service providers available in the data center facility where the cloud infrastructure are supporting your critical applications.

So in a nutshell, if you take into account security, control and availability, the cloud is the way to go especially if you are a web property and are running a Software As a Service. Don't let yesterday's outage make you run back to a mainframe, learn from it, and make your cloud vendor accountable so you do not experience an outage like that with your own applications.