Assessing the impact of continuous change on the storage industry.

The first part of this article appeared in the January issue (Q4 2001) of Storage Inc., the supplement to Computer Technology Review.

On December 6, 2001, executives front Colorado-based data storage companies met at StorageTek's headquarters in Louisville, Colorado. Mark Ferelli, Editor-in-Chief for Computer Technology Review, hosted an illuminating roundtable.

Mark Ferelli: Business continuity can go to issues of legal liability as well. There were legal firms in the World Trade center who lost everything, that did not have even the basic backup strategies in place. They lost contracts, they lost Board of Directors' minutes. And so not only are they in a position to have their doors closed, it is suggestible that their own clients could sue them for malpractice for that kind of inadequate handling of the very important assets that are their documents. It is a scary, scary prospect.

Bill Reed: The dramatic event of 9/11 in terms of waking everybody up to the vulnerability, perhaps distracts us from the much more frequent attacks that data is subjected to through viruses. I've been reading recently the projection of (and I believe the organization is called Surge) forty thousand individual forms of virus attack this year, with a continuing increase next year. One of the byproducts of consolidation is that in effect you can increase vulnerability to that kind of attack simply because you have got more data in one place. If you break into that one place you have access to that much more data to do damage on it. I think that is one area that the storage industry, with its emerging influence in network or storage implementations, has done a very poor job of addressing. It is certainly one of the things that we frequently run into among our customers as we suggest that they tie all of these elements of the storage systems together with either a Fiber Channel or IP networking implementation. The next question is, what about security?

Walt Hinton: The ideal world has a secure file system. But secure file systems need to be not some third-party hybrid thing that everybody has to go and install, it needs to come from people who are building operating systems. Security shouldn't be a bump in the wire; security should be embedded at the OS and in the file system.

Mark Ferelli: Is it happening? Are the court wizards up in Redmond addressing that; are the UNIX gurus and the new class of Linux Luminaries doing things in that area?

Meg Heller: In some respect we have been very much focused on the Solaris environment and building security into that and then also from a storage environment. We acquired LSC with their USF file system, really looking at the whole storage security and how do we build that in and make sure we have the correct links into the Solaris environment. And then we have got to focus on the interoperability piece, because we are in heterogeneous environments--like it or not. We have got to figure out how that is going to work and work with the court jesters at Redmond to figure out how we protect our customer's environment with security.

Walt Hinton: This may be fairly controversial but I do not believe that there should be a security marketplace, period. The fact that there is this billion-dollar plus security industry, the OS manufacturers should be ashamed that it exists. Because it is the wrong place to put it. Think about it, a bump in the wire, well, let's see, all I have to do is get on the other side of that wire and I can invade the operating system, I can invade the file system and do whatever I want. I believe that performing both authentication, encryption, access control, non-repudiation, all of those elements belong, computer to computer, embedded in the operating system.

Mark Ferelli: I've been wondering, as this new creature called storage virtualization begins to evolve, whether virtualization isn't one of the affordability elements. Is storage virtualization going to bring a broader level of affordability to mass storage, and if so, where is it going to play?

Gary Wright: Compaq has been very involved in what we call block-level virtualization and we actually had our first product in 1999. We just announced in October of this year a virtualization in the new high-end array controller and we absolutely believe that virtualization will be the standard in the future. The thing that block-level virtualization can do for customers is to tremendously multiply the productivity of the storage administrator. You've talked about the three things they are doing. [The first is] backup--virtualization addresses that through the snapshot capability, and by the way snapshots do not protect against all types of failure which is why we've had this discussion about tape as an archive medium, and business continuity. The second thing is capacity planning. That's an area where block-level virtualization can really simplify the world for the storage administrator because you are now looking at a pool of storage instead of managing individual volumes. And then the third area was adding storage, and that's where virtualization can make that absolutely transparent, by adding capacity to the pool. So we don't see any choice as we move forward in the future but to offer our customers virtualization.

Jeff Laughlin: I'll take a stab at a definition. Storage virtualization is simply removing the computer's physical, "logical" image of its storage farm. There was always a one-to-one relationship between the host's view of its storage with what actually resided physically out there. My definition would be that we have broken that view.

Mark Ferelli: What does virtualization, either on the block, or the file level, in-band or out-of-band, add to the tape value propositions? Does that enhance the value proposition for tape technology?

Jeff Laughlin: Yes, starting with, I'll call it transparency. From the end user's point of view I don't have to worry about what kind of drive type I have, what kind of library type, in a virtual tape environment. Another one of the value propositions is simply getting more work done for that customer's backup needs to free them up to do more of their company's work.

Juan Rodriguez: We are really looking at levels of HSM which are getting more and more sophisticated. Issues of tape, of sequential access become less and less of an issue. The idea of being able to place the information at the access level puts less and less on performance access, performance of tape, which is good for tape. It makes things much more tape-like, or tape friendly-like, as opposed to some of those fast-action requirements placed on tape and on optical disk which is sometimes ludicrous. So yes, it becomes much more friendly to tape.

Bill Reed: I would agree with that. I also think that the encompassing HSM concepts under the virtualization umbrella in fact help distance us from the black eye that HSM earned itself in the early distributed-computing era. One of the barriers that I perceive right now exists with the backup software vendors and the fact that they are quite frequently wedded to device control, and virtualization has got to insert itself in there in order to break free of the device-specific orientation of a lot of that software right now. I think that the customer benefits from that are enormous. Today literally the success of your backup is limited to one tape drive, if that tape drive happens to be the one that fails, there is virtually no chance of diverting that backup job to another drive without a huge disruption in operation, and very frequently will mean intervention. So the ability to abstract from the individual device and essentially, if there is an issue with one device, you find another one and proceed. There is enormous value in that and great management savings.

Mark Ferelli: How painful is it for the end users to migrate from one technology to another nowadays? I'm specifically thinking of the situation where StorageTek is retiring Redwood support and is migrating their customers to the 9940 platform, which is a high reliability, high availability format and that's great as far as it goes, but getting there can't be half the fun.

Jeff Laughlin: It's not even a quarter of the fun. Bye, bye Redwood. The point is to move from a less reliable technology to a more reliable technology. And obviously there were some issues with data retention over long periods of time residing on helical-based head technologies. I found that we couldn't read tapes that were written by oil-industry people many moons ago.

Juan Rodriguez: I've been in this industry for a long time. I think I started back in 1963, and IBM kept working on tape, half-inch tape. Every time there was a change in the technology from one level to the next, there are always problems. The idea of bringing back tapes of an archive from a long time ago, I think the first tapes that were used in the early fifties were based on celluloid, and the stuff, the oxide flaked off, so it was kind of hard to make that stuff. So I think the issues of technology are less of an issue than the fact that when you are moving data from one space to another there are a million problems, not the least of which is the system technology that is being used to process that information. I've seen big changes at NCAR, when they went from a half inch tape to 3480, or when they went from the first helical which was that terabyte memory, so every one of those changes always, always, creates a lot of headaches.

Mike Harrison: I'd say, from a technology migration standpoint the actual physical moving of the data is nothing. What keeps the customer up at night is the routines that they have embedded into their applications that tie them to the technology. And this cycles us right back to the interoperability discussion, and the fact that we need to be dealing with each other on an open API level so that customers can choose to go to whatever technology base they deem appropriate and not have to worry about writing all of their routines over again for backups or disaster recovery.

Mark Ferelli: As the professional services people have noted, though, people don't know. They have a problem selecting what technology makes sense, either because of lack of technical knowledge or expertise in house, or pure terror. You speak with the National Archivist of the United States, who gave a speech recently. He held up a piece of paper and said, "This is my problem. I've got millions of these things, and how am I going to handle it?" You have never met an organization more terrified blind of technological obsolescence than professional historians. How often do we recommend that people refresh their tapes in order to make sure they have things available?

Juan Rodriguez: That's not the issue. If you look at some of the oldest forms of storage, some of the stone hieroglyphic stuff, the problem is not so much reading the characters but figuring out what they mean. So we are having the same problem. I remember back, I think it was 1988, at Exabyte, we were trying to make our product compatible with Univac, or Unysys, whatever that was at the time. And one of the product specifications was that it was supposed to read back. And UNIVAC forced us to give them a false read backward, basically false We did it all in memory. But the point was that their applications were embedded with operating system commands, which kept re-issuing backwards commands to this drive. It could never work; they could never get rid of it. They started it again, within a month they had a re-package command, which ruined the whole operation. So this issue of legacy issues from applications which are forever old and are never destroyed but are always built on, and these hidden commands that were there somehow or other are the biggest problem.

Jeff Laughlin: Juan is one of our heroes here, and just for his information; we are still supporting government sites that have nine-track. They won't give up. Even though we have offered for free to convert it, there is a comfort zone there.

Steve Lambourne: Realistically most organizations have a cycle where they will write stuff just to make sure it's okay. And if I come to you and say, I've got this new device to replace tape, and it is good for thirty years, your response should be, well so what? I probably won't be able to read that same media in thirty years because you won't have a device to read it. But that is not going to stop the evolution of technology.

ROUNDTABLE PARTICIPANTS

Jeff Laughlin, Director of Strategy, Automated Type Solutions, StorageTek

Robert Sellinger, Chief Technology Officer, Chaparral

Joe Raschke, Business Development Manager, Yotta Yotta

Richard Search, Director of Marketing, Enterprise Storage Group, Compaq

Steve Lambourne, VP Product Marketing, Hitachi Data Systems 

Michael Harrison, Director of Business Alliances, IBM

Walt Hinton, Chief Architect & Senior VP of Design Services, Managed Storage International

Juan Rodriguez, Chief Technologist, Exabyte

Bill Reed, VP of Marketing and Business Development, SpectraLogic

Meg Heller, Director, Storage Services, Sun Microsystems

The last part of this Roundtable will appear in the March issue of CTR.