All the cool companies get high on clusters: high performance, high availability, and high costs. Sometimes, it seems like companies put in clusters because everyone else is doing clusters. Clusters are powerful, and they can do wonders for organizations that really need them. But they are very complicated, and can do more damage than good if they’re not designed and installed correctly. In a poorly configured cluster, a single computer failure can create total data and service loss. In this chapter, we’ll teach you to say no to clusters (unless you live in California and have a prescription from a specialist).
Cluster technology allows a number of computers to act as a single server by sharing resources. Exactly what resources are shared depends on the purpose of the cluster. For high performance, clustered computers can share memory and processing power. For high service availability, clustered computers can mimic one another by replicating data and application code to each machine. RAID can work across a cluster to combine the storage subsystem into a giant network drive. The more you want the cluster to do, the more complicated the configuration and management becomes.
In reality, a cluster is simply a group of independent computers that are combined to work as a system, a redundant system. Once the clustered group is properly
set up, it exists as one single device on a network. After all the talk about RAID in the previous chapter, it would suffice to say that RAID is to hard drives as clustering is to servers. In other words, apply the concept of creating a redundant array of hard discs to a broader level. A redundant array of inexpensive computers you ask? Yes, except for the inexpensive part.
A comparison of clustering to RAID, while providing a friendly analogy, is not very accurate. Major differences exist between both concepts. For example, clustering does not have the hands-off internal management system prevalent in most modern RAID setups. A clustered group is managed by a single system, but that system requires the use of carefully arranged software and hardware.
Clusters can be a cheap way to get supercomputer-scale power. It also can be used as a high availability tool, useful for keeping critical services running 24x7. In some cases it’s even used as an alternative backup system. Statistics show that it’s cheaper and easier than ever to build a cluster. With all these benefits, it’s easy to make a case for clustering in almost any organization.
The above information is the start of a chapter in "Network Security Illustrated," published by McGraw-Hill and available from amazon.com, as well as your local bookstore. The book goes into much greater depth on this topic. To learn more about the book and what it covers, click here.
Below, you'll find links to online resources that supplement this portion of the book.