clustering versus load-balancing

Seems that often times the concept of clustering is often mixed up with that of load-balancing. I thought I’d quickly write something up to help clarify the difference between the two.

Clustering — Making a group of application providers appear as one application provider. A cluster generally consists of 2 servers that share common storage and function as a single, logical unit. The clustering functionality typically is provided by a piece of software (such as Microsoft Cluster Service) that runs on top of the operating system to provide redundancy and failover of the underlying system. Database or Mail servers are usually candidates for clustering.

Load-Balancing — Shifting the computational task of an application across various providers / servers based on some predetermine heuristic. The servers do not share storage or other resources, but instead run their own instances of the load-balanced applications. For example, in a pool of load balanced servers, each server would run a web server (such as Apache or IIS) with an instance of the same Web site, but each would be a separate copy. A load-balancer (either hardware appliance or software that runs on another server) is needed to manage the incoming traffic and balances it out to the servers in the server pool.

If you’d like to read up some more about clustering and load-balancing, here are a few good links:
Network Load-Balancing Blog
Linux Cluster White Paper
Technical Overview of Windows Server 2003 Clustering Services