Compaq Bets On SANs: ENSA Is Their Ace In The Hole.

These are not the days of wine and roses for Compaq Computer Corp. Over the last six months, the Houston-based PC giant has seen its CEO and CFO resign; its profits wane and its stock value plummet; its PC market share continue to erode, and its expensive mergers ruffle feathers and force layoffs. Combine all these ingredients with difficulty in mastering direct distribution and you have the recipe for another difficult half ahead for the PC king.

But fortunes in the computer industry can change as quickly as stock prices, and past problems are often little indicator of future performance. Compaq is betting on several new technologies which it hopes will position the company at the forefront of technical innovation as the millennium draws to a close.

In particular, Compaq is bullish on SANs. As both a storage and an enterprise systems company, Compaq sees huge potential in the demand for high performance storage, backup, and disaster recovery solutions that can be distributed throughout the entire enterprise--without sacrificing speed and reliability. Such is the promise of the SAN. But perhaps even more important, for the company that can achieve the feat, great opportunity exists for simplifying and universalizing storage management. This is the charter for Compaq's ambitious Enterprise Network Storage Architecture (ENSA) strategy, which was announced in December and which the company hopes will be fully functional by by this time next year.

With both ENSA and SANs, Compaq hopes to make storage a more flexible, more reliable, and more accessible shared resource across the enterprise. To learn more about the company's plans for both technologies, I recently spoke with Darren Thomas, vice president of the Multi-Vendor Business Unit of Compaq's Storage Products Division.

JOSHUA: What is Compaq's vision of the SAN as a technology?

DARREN: SANs have generally been defined as storage connected behind servers, and there has been lots of public debate about what is and is not a SAN. In reality, there are many types of SAN, from a redundant pathway with high availability and NSPF to a long-distance connection with a single point of failure. These look different, but are still SANs. With ENSA and SANs, our view is that storage should be a utility on the network, and by utility I include all three types of storage: device to server (direct connect), device to network (NAS) and Multi-device, multi-server (SAN).

JOSHUA: When you say "utility," this does not sound like a storage-centric vision, it sounds like a network-centric vision. Is this the case?

DARREN: By utility, I mean that wherever you put storage, in whatever form it is connected, you will be able to access it through software on the network. This is a vision of automatic, scalable, transparent storage, separated by distance but still in the same virtual pool.

JOSHUA: If SAN-based storage, as a technology, is dependent on the network, why are the large networking companies to hesitant to invest both time and money in SANs?

DARREN: Networking companies want IP [Internet Protocol]. That's their expertise and that's what you see them concentrating on now with new technologies like Voice over IP. SANs use Fibre Channel, which means FCP, and until there is a switch to IP, they see little value they can add. I have talked to the big networking companies, and they simply don't think they can bring value until FC uses IP.

JOSHUA: This is an interesting point, because when I spoke with a 3Com official, he told me that he believes the dollars projected for SANs will go to the disk drive makers, not to the companies providing the SAN network infrastructure. Do you agree?

DARREN: I think to some extent that's true. What you have in a backup SAN is basically a version of the 80/20 rule, with 80 percent of the costs going to the devices, the servers and the arrays, and very little going to the network. I mean, in enterprise backup solution, you have a $70,000 DLT and relatively expensive servers, and for the interface you have maybe some fibre cables and a FC hub, which at the most cost a few thousand dollars. The total cost of the infrastructure is very low, comparatively.

JOSHUA: But doesn't the SAN-both as a technology and as a "movement"-need the networking companies on board, if for nothing else than for exposure and to move storage out of the proverbial basement?

DARREN: I think what you're seeing is the networking expertise being provided by the smaller start-ups, companies like Gadzoox, like the Brocades and the Vixels that have the fibre expertise to offer, and I think they have been very successful. Other companies like Compaq, IBM, and HP who have the resources contribute to the R&D and help with things like marketing.

JOSHUA: Storage is the industry's dirty little secret, the crazy aunt in the attic. Is the SAN the technology that is finally going to put storage on the map and give it mindshare among customers?

DARREN: Believe it or not, we have been seeing a major shift in our customers' views about storage. They are now spending 2-5x the dollars on storage that they are spending on networking technology, and they are requesting quotes on storage separate from servers. The first question they ask is about the OS, then about storage, and then about the server, and in many cases the server decision will be based on their choice of storage. In the past, storage was almost never a strategic decision, and now it is, even to the point of coming before the server. Servers now actually differentiate themselves via storage.

JOSHUA: Is its strong association with Fibre Channel going to hurt the SAN? Where does SCSI fit into the picture?

DARREN: SANs and Fibre Channel are nearly synonymous. SCSI has been and will continue to be a mainstay in the small to medium size business: you can run an inexpensive SCSI network around your office very easily, and there is huge volume in these markets. What we see happening is that FC SANs are going to dominate new installations at the enterprise level, or fibre is going in to consolidate equipment. But the great thing is that the cost to use both protocols is low; it's not like you have to get rid of all your SCSI devices if you want to put in a FC backbone. They can be mixed and interchanged, and we see customers using SCSI internally and fibre externally.

JOSHUA: In what other ways are you seeing FC used?

DARREN: We see FC being used as a direct connect initially, and then later in SANs once the kinks worked out. Interestingly, a SAN is the best way to implement a server cluster, and as a result, clustering is really helping SANs by making servers more reliable.

JOSHUA: Can you talk a bit about Compaq's partnerships in the SAN space?

DARREN: We have partnerships that are intended to expand the uses and reach of storage. With Brocade and CMT, we are developing a product which will transition from FC to ATM, allowing backup and storage to occur hundreds or thousands of miles away, something impossible using fibre alone. We are also working with StorageTek to develop products that would virtually back up systems at high speeds using tape protocols, but writing to media that is faster than tape. And we are working closely with VERI-TAS on SAN standards.

JOSHUA: What's the status of SAN standards in SNIA?

DARREN: I consider SNIA the "Congress" of SAN technology, and it is now in charge of standards. Along with several other companies, Compaq has added more technical people to the working groups so that standards can be hammered out. As a result of our April meeting, SNIA has delivered standards schedules, and we expect them to be adhered to. But the thing you have to realize is, for the most part, the compatibility issues are not in hardware but in the software management pieces.

JOSHUA: There isn't likely to be native, OS-level support for SANs in Windows Server 2000 (NT 5.0). How important an issue is this?

DARREN: If you look at Microsoft historically, they usually wait until the industry hammers out and agrees upon standards and then adds them to the OS.

JOSHUA: Tell me about ENSA. How important is it to the company's overall storage strategy and what can we expect to see in terms of storage technologies?

DARREN: Well, ENSA is really the guiding light around here. It is the leveraging of industry standards for the centralized management of distributed computing.

JOSHUA: [Laughter] OK, that's the boilerplate, but what is it really?

DARREN: That's really what it is. What we foresee is basically virtual storage pooling, which goes back to the idea of storage as a utility on the network. It will allow for data replication over distance, automatically and transparently.

JOSHUA: Should this be viewed as the complete and final integration of DEC's StorageWorks division?

DARREN: This was arrived at by both teams and is the heart and soul of [the Storage Products Division].

JOSHUA: Other than ENSA, what's the status of the DEC acquisition in terms of products and product lines?

DARREN: Well, we will very soon have a single disk drive used in every server, and we are integrating all our array controllers so they use the same code. This is probably the most and best integrated group, technologically, in the company.

JOSHUA: To conclude, can you give me an idea of what the future holds with regard to initiatives like Future I/O in SANs?

DARREN: I can tell you that FIO is still many many quarters (if not years) away, but it basically calls for changes to HBAs, which as you know are really the server's portal to the network. Compaq is pro-industry standard, and we intend to support any standard, from any company, provided that it has a broad base of industry support.