Enterprise storage is said to be in the midst of a revolution: a profound break with the models and paradigms of the past. At least, that's the claim found in virtually every brochure, white paper, or Web site fielded by vendors of storage area networks.
Many SAN vendors go so far as to describe their products as Clayton Christensen-style "disruptive innovations." This much-misused reference to Christensen's book, The innovator's dilemma -- in addition to being an oversimplification of the author's thesis -- implies that the vendor's SAN is so superior to other storage products in the market that its competitors will be forced to imitate it if they ever want to see another profitable quarter.
Also implied in the "disruptive" thesis is a thinly veiled warning to consumers: failure to buy the vendor's product will lead you inexorably toward a technological dead end. Quite possibly, such flawed decision-making will bring about total business ruin as your competitors adopt strategic SANs while your company's technological edge becomes forfeit.
To SAN vendors making such claims, "disruption" is a synonym for "revolution." In this view, they share a contemporary interpretation of revolution with another self-described disruptive innovator, Fidel Castro, who once said, "A revolution is a struggle to the death between the future and the past."
To the SAN vendor, storage area networks are the future; direct-attached storage and network-attached storage are
Adopting modern revolutionary rhetoric, SAN vendors will tell you that they have struggled for the past six years to break the chains that bind your storage peripherals to your servers. Their goal, ostensibly, is to remove a class system that has subjugated storage to the "dictatorship" of the server operating system. They envision a future in which storage exists as its own intelligent and self-maintaining infrastructure without the need for server governance, an infrastructure in which all storage platforms and servers are equal. In effect, the vendor's SAN will free data for use by any application anywhere at any time.
It all sounds pretty good, of course -- until you do your history. The problem with 20th Century revolutions is that they never really changed anything. They just substituted one dictatorial regime for another. For example, the so-called dictatorship of the proletariat in the former Soviet Union never faded away as it was supposed to in Marxist theory. Instead, it became institutionalized, a fixture dedicated to suppressing future (counter-) revolutions. Maybe that's because the modern revolutionaries never really wanted to eliminate class systems, they just wanted to be part of the master class.
So, too, it is with SANs. While most of the advocates of SANs talk a good game about the benefits of utility storage, most will also tell you that the utility infrastructure should be constructed using only their own switches and storage platforms.
Maybe I'm just wrapped up in the 18th Century interpretation of revolution, being that the 4th of July is fast approaching. As part of the celebration of Independence Day at my house, we set aside some time to contemplate the words of the Founding Fathers and, invariably, we find new meanings for their wise sayings within the context of our contemporary lives.
Revolution meant something different in the 18th Century than it did in the 20th. George Washington summarized it as "raising a standard to which the wise and honest could repair." In short, the goal of the revolution was to create something of real value, a foundation that could be modified and improved over time. This is a far cry from the 20th Century revolutionary zeal to bring history to its conclusion once and for all.
Moreover, the intention of the 18th Century revolution was not to drag the masses kicking and screaming toward some historically inevitable end or goal. In the words of John Adams, "the Revolution was effected before the War commenced. [It] was in the minds and hearts of the people; a change in their religious sentiments of their duties and obligations...This radical change in the principles, opinions, sentiments, and affections of the people, was the real American Revolution."
Interpreted and applied (at great pain) to the context of enterprise storage, the thinking of 18th Century revolutionaries might be better interpreted as favoring evolutionary, rather than disruptive, technological innovation. In other words, if storage commoditization under the rubric of a SAN is somehow inevitable, then it should come about on its own, without the need for SAN versus DAS/NAS holy wars. Its technological and business value should be self-evident to consumers and they should vote for the technology with their checkbooks. A battle should ensue only if the consumer is denied what he already determines that he needs.
That's what I call a revolution.
SAN technology ought to be viewed for what it is: an experimental platform still very much in development. The Fibre Channel SAN variant, though not truly a network, has strengths that may earn for it a place, not as a general storage infrastructure, but as a powerful medium for scaling storage behind an intelligent storage server.
What is an intelligent storage server? It is rather like a NAS "head," a thin operating system server wedded to a back-end disk array -- or, just possibly, to a number of platforms arranged in a FC fabric.
Realistically, such an intelligent storage server platform can deliver more than simple NAS or SAN alone. By expanding network protocol support to include both file system-based data access (via NFS, CIFS, HTTP, FTP) and block-based data access (via iSCSI), such a NAS/SAN hybrid platform could provide most of the gains promised by Holy Grail SANs -- non-disruptive scalability, manageability, secure accessibility and intelligence -- but with far less of the pain caused by non-intelligent storage switching and unmitigated vendor infighting.
The latter point is worth a bit more explanation. To a greater extent than FC SANs, hybrid platforms have the potential to make vendor storage platform differentiators a moot point. We are already seeing products emerging that will enable virtually any brand name storage array (or combination of arrays) to be placed behind a hybrid thin server OS. Hitachi Data Systems, Network Appliance, and even EMC have to be concerned by the arrival of the new intelligent storage server heads from Spinnaker Networks, Zerowait, and others that promise "plug and pin" compatibility with their storage array products, but at a fraction of the price. Such a solution is likely to resonate with consumers and help to further reduce all disk storage to commodity status.
Sound far fetched? Not if your remember history. It has happened before: Over a decade ago, EMC displaced IBM DASD in many data centers by promising as-good-or-better performance, together with plug compatibility, and a reduced price. Amdahl used the same strategy to carve out a piece of IBM's mainframe business. Both succeeded in displacing a market hegemony because times were tough, money was tight, and consumers were tired of paying too much for the same technology that someone else could deliver for less money. Those times are upon us again.
So, is such an intelligent storage server platform revolutionary? Not in the 20th Century sense of the term. This is more like 1776-style revolution. Unlike the so-called SAN revolution, consumer mind share doesn't need to be cultivated or cajoled to see the business value of the intelligent storage server. Common sense and cost-savings alone will guide them to the architecture.
About the author: Jon William Toigo has authored hundreds of articles on storage and technology and authors the monthly SearchStorage "Toigo's Take on Storage" expert column. He is also a frequent site contributor on the subjects of storage management, disaster recovery and enterprise storage. Toigo has authored a number of storage books, including The holy grail of data storage management.
This was first published in July 2002