As organizations become more dependent on their information systems, they begin to look for ways to make those systems more resilient. Data center managers have used focused on high-availability architecture for years, but the industry transition to virtual data centers has caused organizations to rethink their high-availability strategies.
More on high-availability architecture
One of the most interesting aspects of the shift to virtual data centers is that even small organizations that may never have used a clustering strategy are now almost forced to deploy some form of high-availability architecture. That's because smaller organizations often treat virtualization as a means to save money through consolidation.
But as organizations consolidate servers, they also consolidate risks. For example, an enterprise may not consider the failure of a single infrastructure server to be catastrophic. But the failure of a server that is hosting three or four virtual infrastructure servers would probably be a different story.
The problem is that unless an organization has a virtualization-specific strategy for its high-availability architecture, the failure of even a single server component can result in a major outage that affects a number of virtual servers. This is why it is so important to implement resiliency in even the smallest of virtual data centers.
Rethinking your high-availability architecture
There are two primary approaches to integrating resiliency into a virtual data center. One approach is to use redundant hardware. The other approach is to achieve resiliency through abstraction. Often these two approaches go hand in hand.
Virtual Data Center e-zine
Redundancy within a virtual data center is similar to the various approaches to high-availability architecture that have been around for years. For example, an organization might use servers that have redundant components -- such as power supplies and network cards -- or they might deploy a clustering strategy. Although there really isn't anything wrong with redundancy, additional hardware can be expensive, and server clusters can be complicated and expensive to deploy and maintain.
Abstraction works by breaking all of the virtual data center's components into pieces and using management software to allocate those pieces. It's also important to know that there are different degrees of abstraction. Some virtual data centers may only perform hardware abstraction -- treating physical servers as a pool of resources -- while others might also abstract applications.
High availability through abstraction won't work without redundant hardware. Unlike more traditional hardware redundancy, though, abstraction-based high-availability approaches often use low-priced commodity hardware as opposed to specialized clustering solutions.
In part 2 of this tip, learn how to create a redundant network architecture.