Should Failures Cast Shadows on Cloud Computing?

The Gmail outage last week left many asking about the viability of cloud computing, at least, according to PC World and other pundits. "Tuesday's Gmail outage was not only an inconvenience it calls into question -- yet again -- the feasibility of present-day cloud computing. One popular prediction is that future computers won't need huge hard drives because all our applications and personal data (photos, videos, documents and e-mail) will exist on remote servers on the Internet (otherwise known as 'cloud computing')."

Every time Twitter goes out, or, in this case, a major free email system goes down, everyone uses the outage as an opportunity to cast shadows on cloud computing. I'm not sure why. In many cases its apples versus oranges, such as Twitter versus Amazon EC2. Also, systems go down, cloud and enterprise, so let's get over that as well.The fact is that cloud computing, thus far, has a pretty good uptime record. Believe me, if Amazon bit the dust for any amount of time you would hear about it. That said, cloud computing providers will indeed go down from time to time, for any number of reasons. However, when compared to your internal systems, they will provide as good or better reliability. I had a client read me the riot act, insisting they not move to the cloud because of the high cost of downtime, but they could not demo their application that day due to the fact that the application's in-house server's hard drive died. Irony.

Cloud computing providers understand that their customers are going to be very sensitive to downtime, and have built distributed failsafe features into their offerings -- well, most of them have. This means that when their primary data center goes down another data center is ready and waiting to pick up the load, typically invisible to the cloud computing consumer. For those times that they have a catastrophic failure, they have procedures in place to get back in service ASAP. Okay, most of them, not all.

So, should you trust the clouds? It's really dependent upon which cloud computing providers you use. While some will have an uptime record that goes well beyond five nines, others won't have the same record of reliability. In these early days of cloud computing, perhaps the best approach is to place less critical applications in the cloud until you figure out the reliability patterns they provide. Once you determine trust, and reliability, you can place more critical applications and data out there.

It's unfair to look at the failure of a social networking site, or public and free e-mail system as an indication of cloud troubles to come. They are typically created and maintained with very different users in mind.Every time Twitter goes out, or, in a recent case, a major free email system like Gmail goes down, people use the outage as an opportunity to cast shadows on cloud computing. I'm not sure why. In many cases its apples versus oranges, such as Twitter versus Amazon EC2...