Big Blue's Great Storage Expectations


VB: Is the storage-management software business turning out to be what the industry projected it would be a year ago or a year-and-a-half ago?

Colby: It is probably everything that people thought it was going to be and then some. [But] I think a lot of us have been slow to deliver on a lot of the promises of what could happen in storage-systems software. It's the clearest point of value ... in terms of what you can deliver for a customer. You still have to deliver an improved cents-per-megabyte or improved reliability or improved functionality at the hardware level. There is still some of that has to go on ... but the most interesting layer is the software layer. That is where you help manage storage administrators' productivity. It gives them the tools they need to be better at what they are supposed to do -- less of an art, more of a science. And that ground, that storage ground, is wide open. There are a lot different functions and features and modules that can come out of there.

VB: When I started writing about storage-management software a couple of years ago, the big catch phrase was "virtualization." Now the terms I hear are "information life-cycle management" or "life-cycle data management," which is the idea of managing storage from cradle to grave. Should customers sit back and wait for the vendors to deliver all of the software pieces, or should they buy pieces now?

Colby: I don't think customers should sit back and wait for everyone to deliver what needs to be delivered. This is going to be a journey we will be in for a long period of time. Customers would be smart to know the roadmap of the vendors they are selecting or the roadmap for the products they are selecting -- what builds off what they are just about to buy, and what might stand alone. I think each vendor is going to have a different philosophy as we go forward. IBM believes that the fervor around open systems needs to be fueled. We need to stay with that and make sure we are pushing on the standards committees to provide those standards. [IBM] will recognize all the major vendors, both hardware and software, in our software offerings. But you should expect us to accommodate for those that we know are dominant or have a high volume or high interest from our customers in the market place. We will modularize our stuff, so when something comes out from Tivoli or something comes out from storage systems that we are putting into the market place, we hope it will be modular enough that the customer can %85 build on a variety of infrastructure that they already started to put into place whether that is Veritas or some EMC. Our hope is that we become the best choice if you want a heterogeneous environment underneath the software layer.

Sponsored post

VB: Is there still a threat out there that the interoperability problem is just going to be transferred from the hardware layer to the software layer?

Colby: It certainly is a place where it is more easily solved than at the hardware layer. You can solve several of the performance and reliability characteristics at the hardware layer. But the overall management of the storage and the data and its placement and its use, I think, can still be done at the software layer.

VB: What is IBM's vision for storage? Do you guys buy into the whole life-cycle data management movement?

Colby: I think in many ways we are driving it. If you look at the concepts of hierarchical storage management, which has been around for a lifetime, [they] are still whole. There is really nothing that different in the concepts. I like to believe that IBM will provide the best solution from a view of all the available technologies, including tape and disk and switches and everything that is in the data path. We will provide the best capability from an automatic standpoint to be able to manage that environment with more intelligence and better algorithms than the next guy. So, yeah, I would say we buy into it.

VB: You say that, in regard to storage-management software, we are on a journey that will take a long time. Give me an estimation of how far out.

Colby: It won't finish. It's just like systems management. Systems management is a journey that started 25 years ago, and I don't see any end to the tools and techniques people will use to further automate the systems environment they are in. I feel the same way about storage. It will just grow over time.

VB: How do you define life-cycle data management? Are all the vendors giving the same definition?

Colby: I don't see it as a definitional thing. When I am answering your questions, I'm thinking of the variety of uses of data, including whether it is highly transactional vs. reference data that is never going to be changed again and is going to be archived at some point. The data will have stages throughout its life that will defined by the use of the data.

VB: So is life-cycle data management a reinvention of hierarchical storage management (HSM)?

Colby: The concepts are roughly the same. Reinvention in a positive way. The concepts of HSM [include] having as much knowledge of the data that you can have ... so a person with less knowledge of the data can make judgements about what to do with the data. Then you have a pretty good intelligence system.

VB: It's my understanding that HSM was part of the mainframe world. So are we trying to reinvent HSM for open systems?

Colby: It was part of the mainframe world as a proprietary system -- that is true. The things that open systems needs are exactly the same things that were needed in mainframe computing 20-some years ago. I don't know if I would use the word "reinvention" or not. In other words, there is no one grabbing the code and rewriting it to work on Unix. That is not what we are doing. But the basic concepts are the same. You have to know what kind of device [the data] is on if it is active; what kind of device it should be on if it is critical; what kind of device it should go to when it is least frequently used but when needed, needed quickly; and what kind of device it will go on when it needs to be archived. Those are all concepts that have been around a long time. Doesn't matter to me whether it's open or Intel or mainframe computing.

VB: So how far along are we in terms of the maturity of the technology when you compare it to HSM? Are we in the infantile stages?

Colby: Infantile is a bit harsh. I would say it can speak, but I'm not sure it can complete sentences yet. Quite frankly, I'm surprised by that. I got out of storage in 1981 and got back in this year. So I'm looking around thinking, "Where is all the stuff I was working on back in the '80s?" It is partially implemented in some of the mainframe devices, and it does not even exist in any of the distributed or small systems devices. And the issues are exactly the same. There is just a lot more of it. The prices are a lot different, but the issues are roughly the same.

VB: So we still have a long way to go?

Colby: Yes, we do. One of the advantages [of being out of storage for a while] is you get to see what the major changes are. When I call on a customer now, there is a vice president of storage management or vice president in charge of data. I'm in shock by that. That used to be a systems programming function that was done, like tuning the system. Now, there are bigger staffs doing storage administration and network connectivity.

VB: So tell me, where is IBM at this point?

Colby: We really do believe that it is software where we should be placing most of our investments right now. The real answer is a software layer that makes things more productive. The vendor that provides the best solutions there will be the leader in the marketplace. And I think IBM is in a very good position. We have a base with Tivoli that is extremely strong. It is well-known and accepted in the market. We brought out the SAN Volume Controller, which is our first Total Storage Virtualization family product. We already have couple of customers installed with it. Virtualization is a concept we have been on for 30-some years in the server world. ... [In addition], we have Storage Tank coming out later this year. It's been announced several times before, and this time it is actually coming out the door. So I really like the position we are in, especially with customers.