In-Depth

Scary Storage

Fall Storage Networking World at exactly the right time, on the eve of Halloween, since it is where you can go to watch the industry eat its young. Big players often use the venue to announce that they are swallowing up smaller firms.

• By Jon William Toigo
• 10/21/2004

Very soon, Halloween night will be upon us. Pint-sized poltergeists and diminutive demons will chase from house to house in search of the ultimate sugar buzz. No car, tree, or home will be safe from pranksters with their egg missiles and toilet paper.

My two-year-old, Isabella, who we call “Izzilla” (after Godzilla) because of her propensity to destroy anything within her reach, has been in training for this date all year. Unlike her older siblings, she may not fully realize that, just before Halloween happens, the family will be whisked off to Fall Storage Networking World in Orlando, FL—an event that we traditionally use as an excuse to take the troops to Disneyworld for a few days.

While my wife, kids, and grandparents enjoy the theme parks, I will dart back and forth to the conference venue to meet with vendors and moderate a couple of panel discussions. From the Izzilla’s perspective, it’s a win-win—one of the few times that dad’s involvement with the storage industry pays dividends. There’s lots of stuff to break at Disney.

From where I’m sitting, Fall SNW happens at exactly the right time, on the eve of Halloween, since it is where you can go to watch the industry eat its young. Big players often use the venue to announce that they are swallowing up smaller firms. It is a truly frightening thing to watch when booth attendants for the newly acquired company ceremoniously (and nervously) strip off their own corporate colors and don the branded golf shirts of their new masters. With more anxiety than they ever experienced from any Jason or Freddie movie, the staffers wonder (1) whether the new shirts will fit and (2) whether their skin will break out in a rash from the one-time competitor’s logo.

Needless to say, a lot of business cards are handed out at SNW, often the advance signal of a burgeoning wave of post-merger job searching and resume writing. (If you like to follow the feeding frenzy of industry consolidation, look for a column in the near future that summarizes announcements and scuttlebutt likely to emanate from this show.)

Mergers and acquisitions aside, “Halloween SNW” was also traditionally the scene of considerable gore and bloodletting, with vendor evangelists fighting mano a mano for that coveted prize: a way to differentiate their increasingly commodity products from each other.

Recently, however, the majority seems to be in a less combative mood. Instead of offering a strategic vision, or even trying to articulate a compelling business case for an innovative product, most brand-name vendors seem more inclined to adopt a blander, “more facilitative” tack. They claim that their key goal these days is sell whatever customers decide that they want, without vocalizing any recommendations of their own. While this low-pressure approach may be welcomed by folks who have tired of the silly wars over in-band versus out-of-band virtualization, or SAN versus NAS, or Fibre Channel versus iSCSI, it is actually a lot scarier and a lot more insidious than the often idiotic and polarizing debates of a couple of years ago.

For one thing, the newfound “customer is king” view is intended to mask the fact that very little of anything new, innovative, or visionary is coming out of the industry leaders. EMC, once a key proponent of the “tape is dead” mantra, is gobbling up tape backup software companies like there is no tomorrow and making deals with ADIC to sell libraries to customers who want them. Meanwhile, longtime tape advocate Quantum is desperately hawking SATA disk boxes as though they invented them.

On the software side, Computer Associates is trying to parlay its considerable enterprise management expertise into a leadership position within the lower realms of storage management, while Veritas looks to be trying to do the same thing, but in reverse: building upon its leadership in the underworld of storage to create a new rep for itself in the loftier realm of enterprise management.

HDS is touting its latest big iron array as a “paradigm shift,” despite the fact that it is a simply a further expression of the “one size fits all” philosophy of mainstream big iron. And, of course, just about everybody is calling their product an ILM solution or a Grid solution or a Utility Storage solution or a Regulatory Compliance solution, though none of them has much to do with information lifecycle, grid computing, utility infrastructure, or corporate governance.

It’s as though none of the major players wants to go out on a limb and actually stand for anything. And in the absence of vendor debates, however substantive or silly, customers have very little guidance for structuring their choices on strategic technology.

In fact, to the extent that there is any advice to the consumer at all, it comes in the form of unsubstantiated platitudes and motherhood considerations. This gives the impression that all storage technology products are just undifferentiated confections in the same chocolate box—which seems to be exactly what the storage bigwigs want us all to believe.

What I long to see (and to prompt where possible) at SNW is a set of good debates. When the vendor says, “It goes without saying that everyone needs a FC SAN,” shouldn’t we hold their feet to the fire to explain why all applications really need this technology? Shouldn’t we make them explain whatever happened to the promised SAN, the heterogeneous and well-managed one that was supposed to deliver downtime avoidance, application-aware provisioning, and non-disruptive scaling? Why do leading vendors now tell us to segregate our fabrics into single initiator/single target zones? Did something go terribly wrong with the “aggregate storage in a SAN” guidance they were giving us just a year or two ago?

When the vendor states, “iSCSI will find a place as an edge technology, feeding a core enterprise SAN,” we need to be asking a bunch of additional questions. Why isn’t iSCSI a core technology by itself? Can’t 98 percent of applications get all the connectivity and throughput that they need from iSCSI SANs? Is the continued preference among three-letter industry leaders for Fibre Channel explained by the fact that there is little wampum to be made from iSCSI, which leverages common LAN interconnect technology?

When the vendor observes that we need a “clustered file system,” we need to be asking why. Why not use a global namespace server or a single instance file system address space solution:? Aren’t these less disruptive and less costly to implement? Or, better yet, what happened to all the talk about object-oriented databases as file system replacements— is there something we should know?

There are clearly a lot of unanswered questions to explore at Halloween SNW. But the current, insidious, “go along to get along” mantra of the industry leaders threatens to dumb down the content of this show. Conspiratorial minds might wonder whether this is deliberate.

Think about it: the most important innovations this year have been from vendors who aren’t part of the SNIA ruling clique. Avamar Technologies, for example, has begun to gain market share with its promise to shrink the space occupied by a file to a fraction of its actual size—18 to 1 under optimal conditions. The underlying algorithm is called commonality factoring and the innovative hosting architecture used for the resulting data is called RAIN. It should be propelling Kevin Daly to undisputed star status, but it’s hardly being discussed.

Revivio’s Time Addressable Storage is another killer app that attacks the expensive big-iron array manufacturers where they live: in their point-in-time mirror split schemes, with all of the overpriced software and expensive disk wastage that mirror splitting creates. Combined with new data replication technologies from Adaptec and Softek, and an increasing assortment of platform agnostic virtualization products, we are seeing the beginnings of a revolution in storage: a capability to deconstruct big-iron arrays, drive down their costs, and to replace them with infrastructure building blocks that have been built with each application and its data characteristics in mind.

The truly scary thing about this Halloween SNW is that no one seems to be asking any questions. I’m heading to Orlando to do what I can to scare up some discussion. Maybe I’ll see some of you there. [email protected]