What I Did Instead of Buying a SAN

Adam Anderson
September 2001

The key concept of a SAN is that storage hardware (i.e., disk and tape devices) is abstracted away from the server and, more importantly, the network operating system (NOS). For instance, it is not unusual to have ample aggregate disk space in an enterprise, but have free blocks in all of the wrong partitions. A SAN solution could remedy this problem by aggregating all storage independently of server and partition conditions. Further complicating matters are the issues of sparing, testing, and purchasing additional disks and arrays for the typical mesh of different hardware platforms. The promise of a SAN is to allow a centrally managed, fault-tolerant, and standardized set of arrays to fulfill all the storage needs of an enterprise. So, instead of learning various RAID management interfaces, sparing many types of drives, and having all your free disk space attached to the wrong server/partition/NOS, SAN technology promises a fully independent and standardized deployment of storage.

Actual SAN solutions, at the price points that fit budgets in typical environments, have not delivered on these promises. Often requiring forklift upgrades and extensive re-engineering of the existing storage model, SAN technology has always been well out of reach of most environments. In situations where SAN technology is not cost-effective, I have used two alternative storage technologies that deliver much of the value of the "ultimate SAN" at reasonable cost and fit well into typical server-centric storage models.