In-Depth

Utility Computing Dark Horse

CA proposes to leverage management expertise for On Demand dominance

• By Stephen Swoyer
• 06/24/2003

Dave Hochhauser, VP of the Unicenter Brand with CA, spoke with us recently about why effective management is the most important ingredient of any utility computing infrastructure. If you think that utility computing is all about hardware and services, Hochhauser says you’ve got another thing coming.

Q. HP, IBM and Sun have all advanced utility computing visions that describe an infrastructure based on a combination of hardware, software, and services. What’s CA’s vision for utility computing?

A. From the On Demand perspective, it’s how do you take the enterprise and make it run similar to a utility. As a utility, we really think that they often deliver this often as a service—as a customer needs it, you plug into a utility and get the computing power or any services that you need.

How do we start running the IT organization to be much more transparent to its users, and transparent obviously to the end user? The answer is that you need to align your IT infrastructure with your business priorities, and that’s part of how we see On Demand.

Everything needs to be based on some kind of business priorities—how do you really align business priorities with your IT infrastructure? How do you automate, how do you make it all as self-managing as possible? Very often, you’re looking at it as just the technology piece and how you get the most out of a hardware box. You know, being able to virtualize an entire computing environment. And while that’s one piece of it, there’s really a lot more to it. It’s really a management problem, which is our area of expertise.

Q. Why On Demand? It’s not exactly a distinct branding, after all.

A. We were going through this and we came up with about 25 different names that we can call the initiative. From our perspective … IBM made [On Demand] an industry term, so why invent a new term? Our unique perspective on it is that we’re in the management business of the On Demand infrastructure.

Q. You’ve stressed the management contributions that you can bring to the table. Do you see any management shortcomings with any of the utility computing initiatives that HP, IBM, and Sun have announced?

A. In some of the cases, yes. We really do believe that the majority of this is really all about management, so when we see shortcomings, we kind of look at it from a management perspective. The whole idea is how to really be more responsive to the end user, so if somebody is looking to do dynamic resource allocation across different servers and computing infrastructures, the real issue is how-when-where are you going to do that, put those controls in place. What triggers them off? What’s the automated policy that allows you to go ahead and say, "Now I need more automated computing capacity."

There are a lot of things around that as well in order to manage that. IBM’s majority of focus is on technology and services, mostly outsourcing services perspective. I think HP, in one sense, is really trying to follow IBM’s image and is also somewhat focused on looking at it from a technology and a services image as well. From our perspective, we’re much more in this game.

Q. But both vendors market established systems and network management products, IBM with its Tivoli Software Group and HP with its OpenView family. These are respected products. Aren’t they capable of addressing the management concerns that you’ve identified?

A. At a high level, while they have management, the whole concept of being more responsive and driving cost control is the management issue, and the focal point that we look from is from a management perspective, and their driving force behind the initiative is not coming from that perspective.

Q. What do you mean?

A. It’s a matter of priorities. It’s critical that we feel that there are two things that we need to be a part of that. We feel that there are customers who are very much looking to get started today and not have to roll out a lot of new technology to get [utility computing]. The other thing is obviously a heterogeneous mix of applications and systems. If you look at an application, it’s rare that an application system runs on just one platform. You have the application on one platform, the database on another platform. A lot of the data might be sitting on a non-mainframe environment, so while a lot of them [HP, IBM, Sun] are looking across a lot of different platforms, we actually are the Switzerland when it comes to that. We definitely have management products across all platforms today, now there’s the idea of progressing and moving forward to becoming more and more what On Demand is.

Q. These vendors will claim the advantage on their own hardware and with their own systems, however. IBM, for example, will point to its Project Eliza hardware technologies—and the ability of its On Demand infrastructure to tap in and exploit hardware autonomic functions—as a selling point, especially among its customers. How do you propose to combat messaging of this kind?

A. They’ll have it on their platforms, but we’ll provide the infrastructure for all of this, as well as support for these things. Something like dynamic reconfiguration: We might be monitoring a Sun box’s utilization, and we may notice that it’s about to exceed a certain threshold, which would drop performance off. We can go ahead and detect that, and then automatically reconfigure that—even before the customer notices that degradation. Sun may have the capability inside its OS to go ahead and re-provision to add the extra blade, but the whole infrastructure for doing that with other platforms isn’t already there, and that’s what we’re going to provide.

Q. Does your vision of On Demand at all preclude your working with rival architectures?

A. In many cases, that’s how we’re going to do this: We’re going to interface with N1, with IBM’s On Demand. As they become native in the OS, then there needs to be a whole management framework on top of that.

So we’re definitely going to tap into [these infrastructures]. We work pretty closely with most of these groups, but we also need to separate the hype from the reality, and we need to know where they really are at any point in time, because that’s where we can interface with them. So we are staying very much in sync with them and we will interface. I can see potentially balancing across different vendors, and maybe plugging into some of their native capabilities.

Q. What about partnerships with these [HP, IBM, Sun] or other vendors?

A. We’re definitely looking at various partnerships. But we’re not just integrating—we’re also looking at acquiring [needed functionality]. Would we look to partner with the hardware players, yeah, but not just the hardware players, and there are other players. Intel is looking at it on a chip level, for example.

Q. If a customer comes to you today and says that they’re ready and willing to implement CA’s version of On Demand utility computing, what can you do for them? Can you give them the infrastructure that you described above?

A. Our first step would be to determine what their understanding is of On Demand. Many companies are a little bit unclear about what On Demand is really all about—and what their goals are. So one of the first steps we would take is really to get organized, take your inventory, understand what you have, it’s an overlooked step in most respects. Think about how to consolidate your servers, how to improve your processes, how to build in a lot of automation. Then you can start proceeding from there and putting in things that might start automatically consolidating servers, or provisioning servers, or deploying them.

Q. And do you have these things [that can automatically inventory, consolidate, provision or deploy servers] today?

A. We have very advanced means of collecting inventory. We’re building the whole idea of resource management and knowing what’s out there. Consolidating servers is a combination of services and technology that we’re working on. Part of it is understanding how these machines are being used; for example, the statistic is that machines are being used less than 25 percent of their capacity. Going forward, there’s more and more automation coming into play, but it’s more of a gradual evolution to get there. We’re not saying that this is something that you can do today. Even if everything was in place to do it today, most companies couldn’t do it, because of the changes involved with people processes and technologies. It’s a gradual evolution to get there.