Tivoli CTO Lays Out Autonomic Computing Strategy

For the past few years, Alan Ganek has led IBM's efforts in autonomic computing, which is designed to automate many tasks associated with system and network management. More recently, Ganek added CTO duties for IBM's Tivoli division. In an interview with Editor in Chief Michael Vizard, Ganek discusses IBM's progress in autonomic computing--a cornerstone of its on-demand computing strategy--and the role he expects Tivoli to play in making that happen.

CRN: IBM has been talking about autonomic computing for a couple of years now. How much progress are you making?

GANEK: We announced last summer an ability that essentially allows for self-modifying chips. It takes the technology in the chip process--which has been there for some time and allows for the testing of different paths--and adds some very innovative, new science that uses a variety of criteria so the chip can be modified. This is in the Power5 chips coming out. You could actually change the circuit design based on environmental attributes that could relate to temperature, voltage and performance aspects. It could sense a different environment that the chip has been placed into and change the circuitry to adapt to that environment. This is a pretty amazing thing and is in the chip technology. It would allow performance problems to be encapsulated and handled in the hardware and not require any intervention.

CRN: What else have you been working on?

GANEK: We found that one of the difficulties in the system is that there is no instrument in a uniform way. Particularly in today's world, applications are really isolated to any one server stack or hardware that the applications are implemented on. But those applications touch a whole set of islands of computational capability. When you go to debug a problem, it is hard sometimes to figure out which island of capability has the problem. The tools tend to be different across all those platforms.

So we went and studied what logs looked like. If all the logs are different and have different formats and different representations, it is almost impossible to correlate them and do that kind of triage. We studied thousands of logs, and almost all of them were doing the same kinds of things, in that there were a dozen or so actions that most of these logs did. We could define a common set of semantics for what the logs represented and then a common format for representing them, which we defined and called a 'common base event.' As soon as we had the idea, we went to the open standards world to try to get this implemented. OASIS is the standards body we have used. It has been developed as the Web Services Distributed Management standard. We have implemented tools that take existing logs and convert it to this format. We have implemented tools that can take logs to a variety of components and bring them together and give you the ability to do correlations, analysis, sequencing and so forth. We've got very good feedback from customers. It is an important step for them in being able to build better automation on top of it. CRN: Where will this come into play as it relates to Tivoli?

GANEK: Some examples would be where Tivoli has introduced a component that does what we call orchestration and provisioning. Looking at the overall environment, it defines workflows that will define what actions to take in what situations, adjusting the overall environment as appropriate. Another example might be identity management. You have an access control so that you can have a secure environment across all of those elements. The control management, identification and monitoring of the environment comes from the Tivoli products. Tivoli provides the monitoring so you know what you are doing in the first place.

As I attack this job, I do it from two dimensions. One is from an autonomic perspective, working with each and every business area in IBM to try to bake in more automated capability and leverage the architectural progress we have made around the various common components that we have built and serviced. Then I look at it from the standpoint of, how do we get the management and control of all these components to work well? That is the Tivoli role.

CRN: What is the overriding goal of autonomic computing?

GANEK: On one hand, we want to encapsulate the autonomic-type capability in the various elements of the system, hardware and software so that they can manage their issues and be self-contained to the extent possible and, as appropriate, communicate to a higher authority when management across the components is of value. This is where Tivoli comes in. Whereas autonomic will cut across all components, the management issues funnel into Tivoli. Tivoli is on a very aggressive improvement path to take advantage of these kinds of capabilities. It gives the management structure that much more to work with. We are upgrading Tivoli to be much better aligned with how people run their IT environments. We're studying how people actually manage these environments and the processes that they use. We're improving Tivoli not only to provide individual products that handle specific tasks, but also to put them together in a way that matches how people manage environments.

CRN: How quickly will that evolve into the ability to better manage business processes?

GANEK: That is happening at a reasonably accelerated pace. It will not happen in one step, but it is happening in pieces as we go. If you look at it from a business-process engineering standpoint, people have looked at the notion of component business models and how you break a business down to what the business processes are. We have been very successful in developing tools for modeling what the business processes are and plugging that into an execution environment. That concept can be applied to any management IT processes. We at IBM can leverage the fact that we are in the business-process and transaction-management business. The on-demand operating environment is allowing for an infrastructure approach that makes the integration of these things much easier.

CRN: Are there any plans to reduce the large number of product SKUs in the Tivoli portfolio?

GANEK: We are certainly focused on the overall product set and how they are consumed and which ones fit together. Simplifying the portfolio is something that we are focusing on.

CRN: One things that competitors like to ding you with is the fact that the Tivoli portfolio consists of a fair number of products that have been acquired and, therefore, have different sets of source code that make integration difficult. What are you doing to overcome that?

GANEK: I think we have a process now. First of all, there is careful selection of what you are buying. When you buy things, it is rarely just to solve an immediate problem. We buy things to fit within our strategy. We are very careful to look at what we have got. Then we work on what the integration is. One of the companies we bought is a small company in Canada called Think Dynamics. We have integrated that to be the centerpiece of this orchestration of provisioning technology. We bought that company in April 2003. We were able to plug that into our production environment to support dynamic application of servers for the U.S. Open by August and September of that year. We were able to reprovision servers from other application pools to the Web site in as little as a couple of minutes. This product was integrated with our provisioning capabilities and our server and storage management capabilities within a few months. It has gone on to be linked to part of our virtualization strategy. I think it has been a pretty remarkable story of being able to take that on, integrate it and get convergence in a hurry.

CRN: At the end of the day, do you think these capabilities will make it easier for solution providers to deliver managed services?

GANEK: Absolutely, particularly in SMB, where people are trying to roll things together. I think you see that in the integrated runtime for SMBs that we built at IBM. By the way, some of the components in that are ones based on some of the autonomic computing components, which allow this integrated logging, for example. If you are a solution provider, you provide some application value based on middleware. But if you have to do a lot of on-site service calls, you go from in the black to in the red. Having a mechanism that strings the products together allows for problem determination and support and a much easier way for remote debugging. That has been a huge winner for folks in that space.

CRN: Why do you think people will spend more time thinking about this space in the years to come?

GANEK: If you go look at the labor that people are spending on in IT environments, it is somewhere in the $600 billion range. This is not counting systems integration, which is probably another couple of hundred billion. The value proposition for managing complexity and system management is going after the problems associated with all of that labor. The value proposition is dramatically larger than the current product opportunity.

You talk to customers, and they will tell you, 'My environments are getting more complex to manage, yet my labor budgets are not increasing.' Therefore, a higher percentage of the labor [they] have is doing management administration, which makes it that much harder for [them] to focus on nonmaintenance operational things--to focus on new business. The biggest check they write is for their people. The productivity of those people, then, is the biggest opportunity. The productivity of those people is exactly what autonomic computing and Tivoli are addressing. You look at it today, and people are spending 60 percent or 70 percent of their budgets on people, and escalating [that]. People want more of their investment to be on new revenue-generating applications, new functions and new differentiation against competitors. They want to spend less on people management and more on competitiveness.