Interoperability: The storage networking challenge

Posted on December 01, 2000

The vendor community has to involve end users in the standards process to avoid mistakes of the past.


Interoperability." In a storage context, the word often conjures up fear, uncertainty, and doubt. Storage networking interoperability poses an exceedingly complex set of issues involving literally hundreds of vendors, many standards bodies and initiatives, security issues that cut across multiple disciplines, and of course, end users.

To help sort things out in my own mind, I've grouped the issues under two headings: technology and process. The technology issues are things like what works with what, or what will work with what and when. Let's put those aside for a minute and focus on process: How are interoperability standards being brought to market, and what is the user community's perception of that process?

Last year, I appeared on a webcast to 2,000 or so Netizens, explaining the virtues of storage virtualization and talking about how storage area networks (SANs) could get us to some of those promised-land benefits. I believed-and still do-that the major hardware interoperability issues had been worked out, and I was recommending SAN implementation to all who were thinking about storage consolidation.

I continued on this track until this summer, when I received a call for a white paper on interoperability. The person asking for the paper was clearly of the opinion that interoperability was a compelling enough issue to print 250,000 copies. And so, somewhat conflicted, I researched the issue in detail, talking to heads of standards bodies and chairs of working groups, tracking down proposals, and following other leads.

I then inventoried all of the applicable standards. I first noticed that there is no lack of storage networking standards proposals. If anything, there may be too many, forcing standards bodies to decide which see the light of day first and which others come later and possibly never. I also noticed that interoperability was a two-stage process, from formulation in the first stage, to consistent implementation among vendors in the second stage. I finally concluded that there were enough "stage two" standards available to deploy SANs, even in production environments.

"Red-herring" issue

In the end, I basically came to the same conclusion I had when I started. So, when a reporter called who was doing a story on storage interoperability, or rather the lack thereof, I was comfortable calling it a "red-herring" issue. There were more compelling impediments to SAN adoption such as high cost and staff inexperience. I told the reporter that interoperability was no longer the main obstacle to SAN deployment. No matter. The article appeared, and like many I have seen, it focused on users who try to build storage networks, only to cultivate storage "not-works."

Again, I questioned myself. Is the disbelief in SAN technology so widespread? If so, could I see it in the vendor shipment numbers? I looked, but the numbers told a different story. Major SAN vendors were reporting meteoric growth in SAN-related sales. Financial analysts were also very optimistic. So, with so many users buying SANs and with the financial community generally optimistic about the future of SAN technology, why are SANs getting such a bad rap in the press?

The answer came in a call from a vendor client who was about to embark on a campaign to call its major competitor's SAN strategy "proprietary" and to publicly accuse the competitor of dragging its feet on standards. While dissuading my client in no uncertain terms from this course of action, I saw firsthand what users see: the fractiousness within the vendor community that leads to end users' negative perceptions about interoperability and feelings of despair for its future.

To verify this notion, I called a number of Storage Service Provider (SSP) CTOs and asked them if the word "interoperability" engendered positive or negative feelings, and if the perception of the vendor community was one of cooperation or discord. In each case, the response was negative. One CTO described interoperability as a "dark cloud" that hangs over the storage industry.

I have come to the conclusion that the word "interoperability" means different things to different people. The trade press often refers to the problem as a "lack of standards," yet standards proposals proliferate like weeds. What the industry really needs is a verifiably consistent implementation of those standards by vendors. And when we think of the vendors involved with the process of bringing open standards to the marketplace, do we identify the problem as interoperability between vendors when we really mean a lack of co-operability?

Let's now turn to users' perceptions of the interoperability process. I recently attended a storage-oriented user conference, Veritas Vision 2000. On the first-day agenda was a free-wheeling panel discussion among vendors from various disciplines, including storage hardware and software, databases, and operating systems. The audience of 2,000 became actively and vocally involved when the discussion turned to interoperability. One question from the audience that I believe typifies user sentiment on this issue was, "When are you guys going to get together and make all this stuff work together?" The audience applauded.

User concerns

Yet, it is hard to find this sentiment in user-opinion surveys. Two surveys published within the last six months show that high cost and lack of staff resources are the leading concerns among users. Interoperability appears well down the list. Nevertheless, there is a direct relationship among cost, staffing issues, and interoperability. To avoid the potential headaches when implementing a SAN, users are often advised to go to a single "trusted source" for fully integrated and operational solutions. Users naturally equate "trusted source" to a single vendor's solution, which in turn, equates to premium dollars paid to make the interoperability issues go away. I believe that, as a result, users feel caught in a vise: fear of SAN failure made all the more vexing by staff inexperience with a new technology on one side, with the requirement to pay a premium price to dispel the fear and outsource the experience on the other. When this happens, users look for someone to blame, and some easy targets appeared before them at Veritas Vision 2000.

Users also look for ways to extricate themselves from the vise-alternatives that I believe threaten the bottom lines of the Fibre Channel storage community. What are the alternatives? First is the storage-as-a-service alternative offered by storage service providers (SSPs). Users have traditionally bought storage infrastructure directly from vendors. What happens if users turn in increasing numbers to service providers? Premium dollars will go to the service providers, accelerating the commoditization of storage infrastructure.

Internet Protocol (IP) storage is another alternative for users, now hyped as nirvana in ways that are painfully reminiscent of the Fibre Channel SAN hype that is now a source of irritation among users. IP is ubiquitous, understood, and best of all, low-cost. But, will those attributes translate themselves to IP-based storage networking? That remains to be seen. And what of the IP storage interoperability issues? Are they all behind us? In fact, the problems for the IP storage community could just be starting. InfiniBand, too, is looming large on the horizon as an alternative to both Fibre Channel and IP.

Plug and play

To forestall intrusion of the alternatives, the Fibre Channel community needs to show a different face to users. It appears to rely solely on interoperability labs to show technical interoperability, while ignoring the absence of co-operability. Users need to experience the positive side of interoperability: plug and play. They need to see vendors cooperating in order to believe that open standards will be implemented consistently. The image of fractious vendors managing the standards process via dueling standards organizations must be dispelled. To do this, standards organizations, specifically the Fibre Alliance and the Storage Networking Industry Association (SNIA), need to directly involve users in the standards process. There are no users, for example, on the SNIA board of directors for 2001. Nor are users directly involved at the committee level. Involving users at the board and working committee levels may seem a radical thought to those already participating, but the time has come to give it serious consideration. In the words of one CIO heard recently, "Absolute participation by users in the standards process is critical."

Users have no direct control over setting the storage networking standards agenda. Clearly, they are not comfortable with the process as it is now practiced, and they fear that the strongest vendors will ultimately take control over the agenda. By more directly involving users, standards organizations now controlled by vendors would send a message that they hear users' concerns and frustrations, signal a willingness to end the divisiveness, and order the agenda in a way that helps users rather than vendors.

John Webster is a senior analyst with Illuminata (www.illuminata.com), a research and consulting firm in Nashua, NH.

Click here to enlarge image

Senior analyst,