SAN Vendors Cooperate? Ha!
If SANs followed open management standards, you could say goodbye to Veritas, Tivoli and Legato, for starters. Relax—it isn't happening anytime soon. Instead, litigation is the order of the day.
Ever wonder what would happen if a truly open, standards-based, highly manageable and intelligent Storage Area Network (SAN) appeared in the market? While the idea is a wishful dream for enterprise managers, it's a nightmare for storage vendors.
If a real SAN were to appear on the scene, of course, there would be nothing left to differentiate EMC Symmetrix or IBM Shark arrays from, say, those cheap platforms built out of commodity drives by some local yokel working from his garage. Disk arrays would become just a bunch of disks (JBODs) managed by a "fat" controller called a "virtualization engine."
Moreover, such a SAN, by definition, would have to be built on a real network protocol like TCP/IP rather than on a serial storage interconnect and network "wannabe" like Fibre Channel.
Assuming that TCP/IP became the plumbing of a SAN, then the host bus adapter (HBA) industry would go belly-up. Why would you need an HBA if you could do storage I/O across the same network interconnect you used for your LAN? A network interface card (NIC), enabled by a TCP stack Offload Engine (TOE) chip would provide all the storage connectivity you needed.
If SANs were enabled by an open, any-to-any management standard such as Common Information Model (CIM), you could wave goodbye to Veritas, Tivoli, Legato, et al. You wouldn't need them anymore. EMC's WideSky would float off into the ether.
In fact, if a real SAN appeared, a lot of companies would need to be de-listed from major stock exchanges. There would be even more overqualified people applying for jobs at Burger King, McDonalds and Jack-in-the-Box.
It's a scary series of thoughtsbut don't let it give you insomnia right now. Arriving at such a platform would require several things that aren't likely to happen anytime soon.
Technical issues aside (and there are many), the development of a true SAN would require either:
- The overnight success of an upstart solution provider, which would require the support of a lot of clear-thinking purchase decisions by Fortune 500 IT departments, or
- An unprecedented level of cooperation among current storage industry playersa level that they haven't demonstrated till now.
The first scenario is unlikely. The business plans of most start-up storage players these days have three key milestones: Establish name, attract interest of established storage vendor, sell company. If a start-up does introduce a technology innovation that it doesn't intend to sell off, it must cope with an array of forces aligned against it.
For one thing, getting into the game requires analyst endorsement. In this effort, money talks, just as it does when lobbying for an Oscar in the film industry or getting a bill through the U.S. Congress.
Second, the technology must cultivate a following among those conservative buyers in Big IT Departments who generally know little about storage, have little time for or interest in learning new things, and who tend to purchase technology on a checklist basis.
Third, the newbie will require reseller channel partners who can put in the "face time" with their customers in order to sell new stuff. To interest channels, margins must be high so that returns are high. This raises the price tag on the new technology and reduces profits realized by the innovator.
The second possibilitythat unprecedented vendor cooperation will bring about a real SANseems even less likely.
Cutthroat competition in the present-day SAN market (the Fibre Channel variant) has already delivered non-interoperable fabrics and proprietary solutions based on single-vendor products. Recently, this situation has worsened through a spate of lawsuits as vendors turn litigious.
Here's the litigation to date:
May 2001: Storage Computer sued Hitachi Data Systems, Seagate Technology and XIOtech Corporation for infringing on its patented RAID 7 technology. Storage Computer said that the offenders had integrated its patented RAID 7 technologya variant on the open RAID standards articulated by UC Berkeley authors in 1987into their products.
September 2001: Crossroads Systems, a vendor of storage routers, won its case against Chapparal Networks over alleged infringement of its patent covering LUN zoning technology.
A similar case against competitor Pathlight Technology also produced a win when Advanced Digital Information Corporation (ADIC) acquired Pathlight and elected to settle the pending Crossroads lawsuit for $15 million.
Despite these wins, the company's first quarter 2002 earnings tallied up to $6.4 million losspossibly a testimony to the slice taken by attorneys in contingency-based lawsuits.
October 2001: Media manufacturer Imation sued Quantum Corporation, maker of the industry dominant DLTape and SuperDLTape products, for operating a "backup media cartel."
Quantum countered that Imation, which had been working to qualify as a media provider for Quantum-technology drives, abandoned its qualification efforts and was preparing to sell unqualified media to Quantum customers. Contrary to Imation claims, Quantum argued that other providers who have complied with its media qualification program are now delivering DLTape media to market.
Quantum told its customers that it would invalidate their warranties if they use unqualified tape media.
February 2002: McData sues Brocade Communications Systems over an alleged violation of its patent No. 6,233,236, which covers aspects of measuring traffic within a switch.
The technology relates to eliminating bottlenecks that can slow data transfer within the switch. McData says Brocade uses the technology in its new large switch, which is the first product from industry leader Brocade that encroaches on McData's turf: The big switch market.
Brocade argues that McData is crying sour grapes and is trying to stifle competition and innovation.
McData argues that it has many more patents than Brocade and points out the paltry patent record of its competitor covers unimportant things like rack enclosures.
April 2002: While this one hasn't yet produced a lawsuit, the breakup of the working relationship between Cisco Systems and Brocade Communications Systems was accompanied by all the acrimony of a Hollywood divorce.
Cisco and Brocade became odd bedfellows in late 2000 when a key customer demanded they work together to enable Brocade-based Fibre Channel SAN "islands" to be interconnected across a Cisco-based IP network.
The companies quickly worked out a tunneling protocol, Fibre Channel over IP (FCIP), that was to be implemented on a blade of a Cisco Catalyst switch to enable SAN interconnection.
Cisco cried foul when Brocade engineered the solution for compatibility only with Brocade switches rather than all Fibre Channel switches. Each company has gone its own way, but neither was willing to give up the last name "Systems."
April 2002: EMC launched a substantial lawsuit against Tokyo-based competitor Hitachi Data Systems (HDS), claiming that the products Hitachi Open Remote Copy and Hitachi Open Asynchronous Remote Copy infringe on EMC's Symmetrix Remote Data Facility (SRDF), and that Hitachi's ShadowImage infringes on EMC's TimeFinder.
EMC has turned not only to the U.S. District Court, but also to the International Trade Commission, which has the power to expedite a decision to block infringing products from entering the United States.
In press interviews, EMC said that it has tried to resolve the situation with Hitachi in an amicable fashion for the past four years, and initiated legal action only as a last resort.
On the other hand, EMC's revenue in the fourth quarter of 2001 was down significantly, to $1.51 billion, compared with $2.62 billion in the same quarter of 2000. In part, the figures reflected increased competition from HDS.
Analysts said that, while lawsuits in the storage industry were becoming increasingly common, this one might give prospective customers pause before buying HDS storage products, and that might benefit EMC's bottom line.
As you can see, vendor cooperation isn't any more forthcoming in a bear market than it was in a bull marketin storage hardware or software.