Somehow, a perception exists that a cloud provides a certain level of redundancy by default. However, make no mistake. Redundancy is not inherent.
Admittedly, individual hardware and software components have some redundancy built in. However, those capabilities do not eliminate the need for a redundant cloud any more than safe cars eliminate the need for speed limits, traffic lights, divided highways and the rules-of-the-road.
For many cloud providers, especially consumer cloud providers, the only redundancy offered is to make physical copies of the data—and many customers do not use even that minimal level of recovery. These clouds were not built with redundancy in mind. They lack the automation, monitoring and procedures to provide clients with an environment that can anticipate, react and recover from component failures. Such clouds are cost effective only if your business, employees and/or customers can tolerate the occasional complete loss of service.
The hallmark of an enterprise clouds is the redundancy it offers. Redundancy exists throughout between the infrastructure layers to ensure high-availability. For example, a failover process detects application hangs and interruptions so corrective action takes place quicker. Monitoring tools ensure no single points of failure develops, and specially-built automation handles error conditions when a problem does occurs, obviating the need for human intervention. This type of automation is particularly important because human interaction comes only after some level of damage is evident.
It is cloud vendor’s responsibility to design and build redundancy into the cloud, and the expertise, staff, time and investment it requires is substantial. Patches and piecemeal solutions added over time do not render the same strong results as redundancy baked-in from the beginning.
Is recovery of stored data enough redundancy for your applications?
Download SunGard’s white paper, “The Real Value of Cloud Computing.”