In this column, I often discuss Storage Area Networks (SANs) and the way their capabilities will eventually change the way networked computer systems are built. However, the reality of SAN implementation has a darker side. Over the past few weeks, several people have said that they've heard SANs don't work. They aren't practitioners—because practitioners know that SANs do work—provided you purchase your SAN from a limited number of vendors who cooperate with each other. Only if your SAN is a storage system attached to a switch—such as Dell's—is your SAN going to work virtually problem free 100 percent of the time.
The question of whether SANs are anything more than a marketing idea to sell more disk systems came into sharp focus as I sat in the audience at Veritas Vision2000, listening to Dataquest Chief Analyst Carolyn ("Numbers") DiCenzo and a distinguished panel of storage technologists including members from Compaq, EMC, Hewlett-Packard (HP), IBM, Sun, Oracle, and Microsoft. Peter Levine, a Veritas executive vice president, told the group that his company had spent more than $100 million building an interoperability lab to test its software with leading industry equipment. According to Levine, technicians had a devil of a time getting different storage systems to work with one another. Invariably, Levine said, when a customer complains about Veritas software, the problem turns out to be a hardware element that has been altered. Levine offered to host an industry conference to solve the storage interoperability problem. The man sounded exasperated—even though the people on the stage represented close partners.
Interoperability has become the rallying call for a number of vendors, who use the password "open" to describe what they offer. At the recent Storage Networking Conference in Orlando, Florida, IBM's Linda Sanford said that the lack of open standards would damage this industry segment. That concern is what drives the Compaq/IBM deal, according to Mark Lewis, vice president of Compaq Storage, and the IBM panelist.
The discussion of openness was largely directed at EMC, although James Rothnie, EMC's senior vice president of product management, said that Symmetrix is the most open of platforms. In fact, Symmetrix probably connects to more OSs and server types than any other large storage vendor in the industry—and most management frameworks support it. However, EMC isn't exactly open when it comes to working with the other large storage hardware vendors. As the industry leader, maybe the company doesn't have to be.
So, is being "open" really important? It becomes so when your customers stop buying your hardware. That hasn't happened yet to EMC. However, as more organizations implement SANs, legacy issues will increasingly plague all storage system vendors. EMC is aware of potential problems. Ben Fathi, general manager for the Microsoft Global Partners Enterprise and Partner Group, for example, complimented EMC's work with Microsoft on the new storage APIs being built into future Windows versions. Fathi noted that EMC was instrumental in this work, even though its eventual implementation will probably cut into EMC's TimeFinder sales.
The Veritas vendor panel's consensus was that a year from now most of these interoperability issues will be behind us—which makes me glad that I don't work in a storage lab. That timing is why 2000 is the Year of the SANs (of course, so were 1999 and 1998).