Q&A: IBM's LeBlanc on WebSphere plans

Big Blue GM talks SOAs, integration plans

As general manager of IBM's application and integration middleware division, IBM veteran Robert LeBlanc has charge of the key technical pieces corporate users need to carry out enterprise-wide integration strategies and other complex strategies such as SOAs (service-oriented architectures). He has the strategic responsibility for the company's line WebSphere application servers, transaction and messaging systems, pervasive computing software, and business integration solutions.

LeBlanc spoke with Editor at Large Ed Scannell about a wide range of topics including his to-do list for WebSphere and related products in 2005, the state of SOAs among corporate IT shops, and how he regards the strategic role of ESB (Enterprise Service Bus).

LeBlanc joined IBM Canada in 1981 as a systems programmer trainee working in the company's Toronto Labs. He has held a number of technical positions in those labs as well as in IBM's Software Group over the past 24 years. With the National Hockey League on strike since last fall, LeBlanc, an avid hockey fan, finds he is getting a lot of work done in his new job. He claims he does not to miss the game given his current workload. We don't believe him.

IW: What are the top two things you want to accomplish with WebSphere in 2005?

LeBlanc: Well, we are focused on a couple of things. We have the next generation of our integration middleware we will deliver in 2005, and we will continue to focus on the integration of all the pieces.

IW: Are you talking about delivering new software to glue existing products together or simply combining products within the WebSphere family?

LeBlanc: A combination of both. Web services definitions and standards continue to evolve and so there is lot more for us to do in the area of Web services. The J2EE standard has settled down quite a bit so there is less work to do in that space, which has freed up my engineering team to really focus on the Web services standards. We participate in every single standard and chair quite a few, so we have a good sense of what is coming down the pike. So, in the next iteration of the WebSphere family, you will see more and more support for open standards.

IW: Well, which ones?

LeBlanc: Many people are starting to pay more attention now to transaction and security. We have some basic support in the [WebSphere] product already but those standards continue to be enhanced.

IW: Is this what you hope to accomplish with the upcoming version of WebSphere, which I believe is code-named Pyxis?

LeBlanc: I never heard it called that. I don't worry about code names.

IW: No, but journalists do. Will it be called WebSphere 7?

LeBlanc: In all likelihood it will be called WebSphere 7. It will be a continued evolution of the set of capabilities we have in the WebSphere platform. We will continue to enhance things like the whole area of systems management. There are a whole lot of things coming in the application development space and tooling space. So when we look at this, we try to look at it holistically. It is not just WebSphere but everything that surrounds it ranging from the systems management tools like Tivoli to the Rational development tools.

IW: So we can look forward to even tighter ties between WebSphere and Tivoli over the next year? Specifically how?

LeBlanc: Closer ties in terms of a set of capabilities and manageability. But that really speaks to our overall software strategy in general, which is to tightly integrate the capability and reuse most of the capability we have.

IW: Project Vela, (WebSphere Application Server 6.0 delivered in December, 2004) was the first step toward the componentization of WebSphere. How are you hoping to get users pumped up about componentization, which has been talked about for what seems a decade?

LeBlanc: Well, there are two things around componentization that make WebSphere that much easier to embed in other products and to use in different ways. From a user perspective, the real value comes in the enhanced capability that we are providing in the Vela release around Web services capability. Things like JMS messaging. When we delivered WebSphere 6.0, we also announced WebSphere Extended Deployment. With this product we are starting to add on top of WebSphere the kind of capability for quality of service for things like clustering and management. I can now add a server to a WebSphere cluster while it is running, for instance. I can start to get some of the system's scale-out capability built right into the platform. And that allows WebSphere to play an even bigger and broader role. We have 18 pending patents on that product. We took a lot of concepts usually associated with big (mainframe-based) SysPlex concepts and brought them down to the distributed WebSphere world.

IW: Is some of the idea behind componentization to take the heart of WebSphere and plunk it into other products, either yours or those of other ISVs, thereby making integration easier among them? Where can you take that idea?
LeBlanc:
That describes it pretty well, and yes it can offer the ability for them to integrate with that product. You can think of almost anything, then, as a WebSphere application. The pieces of the other products can then communicate very easily with each other. You would have common data constructs underneath, and you do not have to do all these transformations. This can help performance and scalability, which are important to users.

IW: Do you have any notable ISVs interested in a deal?

LeBlanc: Well, there was the agreement we had with PeopleSoft until Oracle acquired them. That was exactly the whole basis of PeopleSoft deal we had. People thought we did that deal in order to keep Oracle out of PeopleSoft. It wasn't. We had been working with PeopleSoft for a number of years. What ISVs are starting to recognize is, the value they can provide is not in the plumbing but in the application. We think we have pretty robust plumbing and we are not in the application game. Internally we are doing that very well. When I took over Tivoli, it sat on its own infrastructure. Now, three years later, 80 percent of Tivoli products sit on top of WebSphere. We are talking to a lot of ISVs looking to do the same thing.

IW: So what does the Oracle acquisition do to your deal with PeopleSoft?

LeBlanc: They (Oracle) have to decide what they want to do. Obviously with the deal we signed, there are provisions for changing control etc. Oracle has a very basic app server. They have to decide if we can provide them value. We are fierce competitors but in another sense if they want to be built on the most robust technology, then we think that is WebSphere.

IW: One interesting part of the Oracle-PeopleSoft deal is Larry Ellison (Oracle chairman and CEO) cozying up to IBM, after years of antagonizing you. How do you feel about that?

LeBlanc: Customers still want to have choice. If Oracle narrows down what you can do and how you can do it, then they will alienate customers and will lose them very quickly. So the integration of the verticals that has been happening … it is a two-sided sword. Integration is good, but it also takes away some of the choice and users want both. I think the industry understands that users value having choice and being open. We built our whole business around those concepts and have done really well. So what you are seeing is Larry's realization of this, and he probably heard it from customers.

IW: What do you expect in terms of further consolidation among the larger players in the enterprise software market? And if there is considerably more coming, will that make your life easier or harder?

LeBlanc: To be honest, it depends on how you look at it. You can say it is a validation of the IBM strategy because what they are all doing through merging is bringing together all these sets of capabilities. They have to integrate companies to get that same breadth and integration we have internally. I don't need to take a storage management company and combine it with a systems management company or an app server company and glue it to a business integration company. They are reacting to customers' desire to partner with vendors who are broad providers and that can help them with the level of integration. Gone are the days of users picking out the point product because then they have to become the integrator. We are seeing users say look, 'I want products that are very competitive, but you know what, if they are well integrated, that offers as much value as does function A on Product B.'

IW: But so few of these bigger mergers work out well, which ultimately puts users' hopes of integrating products from those vendors at risk, doesn't it?

LeBlanc: And that is why you want to do it with open standards. With open standards you have the ability to unplug one piece and plug another one in. That is why we are so maniacally focused on everything we do around open standards whether it be J2EE or even, on the tool side, Eclipse. People used to think of integration and best-of-breed as an oxymoron. Well, you know what? If you do it on an open standards basis, you can realize both benefits. And that is what we are trying to drive.

IW: So with IT budgets loosening up a little over the past year, are you seeing more IT integration projects that were shelved now getting dusted off?

LeBlanc: Integration projects were always very much in corporate IT budgets. If anything, we have probably seen a little bit of acceleration there. If there is even a little bit of money, integration projects have typically been one of the areas that users target. Now, though, users who were not even thinking about integration are giving it very serious consideration because they understand how it can affect their business. What might be new is that, for the first time, we are seeing this being driven not from the technology side but more from the business side.

IW: Are you seeing any shifting within the priorities IT shops have for the way they approach integration projects?

LeBlanc: Integration projects tend to be longer-term discussions; they are not something you decide to do when you wake up on a given morning. It is not like selling a word processor. Users are really starting to think about their next generation of architecture. So they think about that and they step back and say, 'what do I want to be the enduring architecture,' and lately it tends to swirl around this whole service-oriented architecture.

IW: Are corporate IT shops becoming more enlightened about what they want to do to establish a service-oriented architecture? If so, how are they typically starting their move to do that?

LeBlanc: It all depends on the customer and the industry. We find the financial services industry tends to be a good leading indicator of what is going on. They are clearly looking at it from the perspective of 'what is the overarching architecture that I want to build my next generation of applications and systems to in order to that get the best value?'. They are really thinking that through. And most seem to be very practical about it. They say, 'okay, I know where I want to be longer term, so let's start with the smaller integration projects today,' knowing that they will fit into the longer term set of architecture decisions they want to make.

IW: Roughly, how many IT shops that you talk to have decided they want to move to an SOA? It is a long-term decision that some tell me can take up to five years from start to finish.

LeBlanc: It is a long journey no doubt, but not one that necessarily takes five years. With our approach we can provide a lot of capability that hooks into your existing systems now with third-party apps such as those from SAP or Siebel. We can also help them hook in their CICS and DB2 applications. Now, if you're asking how long it takes to permeate through everything they do well, that is over multiple years.  A survey among CIOs taken by Merrill Lynch last fall said that about 70 percent of all customers had some SOA-related project in active engagement.

IW: That percentage is a little surprising based on some of the conversations I have with most corporate IT executives.

LeBlanc: Well, this takes [into account] any project, from something in their architecture team to being in deployment. If you are talking actual deployment, the number is substantially lower. But active work going on from an architecture perspective, that 70 percent figure seems reasonable to me.

IW: IBM in the past has not really seemed too hot on ESBs. What is your feeling about the technology?

LeBlanc: If you step back and look at what an ESB is, it really is a design pattern around messaging as well as a set of services around messaging such as mediation, quality of service, and things of that nature. If you think about it as being that, then you can say we have been building technology around that concept for a number of years. There are some people in the industry who are treating this like it is brand new and like it is this thing -- and ESB is a thing. We look at it more like it is a set of patterns in putting together an SOA. You can implement an ESB with the technology we already have.

IW: Do you see ESBs as enablers for people who are just starting to piece together an SOA?

1 2 Page 1
Page 1 of 2
How to choose a low-code development platform