Introduction: SANs Will Raid SCSI Turf, Eventually

With the huge marketing emphasis on SAN technology, it may seem as if traditional SCSI/RAID subsystems -- the storage industry’s poster child of the early and mid-1990s -- are being sentenced to an early death. The truth is that there’s probably nothing further from the truth.

Although SANs are claiming the majority of marketing mindshare in the industry, traditional RAID technology and SCSI interfaces still have a place. In fact, given the immaturity of many SAN components for many applications, SCSI/RAID or SCSI/JBOD technology is the safer bet -- and probably will stay that way for some time to come.

One of the chief problems SANs face is that the underlying technology being used to build them is maturing slower than anybody expected. Peel apart a SAN and you have Fibre Channel technology as the primary backbone. Fibre Channel has been the next great technology for an eternity now. And were it not for its use as a storage area network backbone, it would be struggling to find a widespread-enough deployment to drive costs down to affordable levels.

I sat through one of the first press conferences that trumpeted the impending arrival of Fibre Channel technology, held by IBM Corp. and Ancor Communications -- about six years ago. While the promise for great performance has come true, the expectations for widespread adoption were far less accurate.

Another significant SAN problem today is the interoperability among products from different vendors. As a result, we’re seeing preconfigured, pretested end-to-end solutions offered by companies that promise a guaranteed working configuration. These are safe, viable solutions -- if you’re willing to buy all your components from a single vendors.

But despite SAN’s "wave of the future" image, customer wins that get promoted by the vendor community still frequently highlight SCSI solutions. For example, Box Hill Systems Corp., one of the first companies to deliver a Fibre Channel storage subsystem, just landed a $2.3 million chunk of business from Lockheed Martin related to the Data Capture Systems 2000 automated check-in system, being configured for the 2000 Census. Lockheed is going with a storage network based on SCSI technology.

There’s another dimension to consider: the fierce nature of the storage industry. Selling storage is a highly competitive business with products that, in many cases, offer similar features. As a result, storage vendors are fast to capitalize on the slightest marketing opportunity to get a leg up on the competition. At the moment, the opportunity is spelled SAN.

The reality is that for most users storage investments take years to amortize. In many cases an end user will eventually bring a new storage system in and then stage older technology along to another, less-demanding application. If you have an application that will truly benefit from installing a SAN, by all means, go ahead. There is SAN technology today that may solve your problems -- just make sure you do your interoperability due diligence before sending in an order. If you’re not ready for SANs, don’t let marketing pressure change an otherwise sound decision.