John Swainson is general manager of IBM's Application & Integration Middleware Division - a role that gives him responsibility for middleware technologies including Web application servers, e-commerce servers, transaction systems and application development tools. Swainson recently spoke with Network World Senior Writer Ann Bednarz about IBM's plans for integrating the IBM family of software products around a core middleware base, including WebSphere as a transactional run-time environment, MQSeries as the secure messaging infrastructure and IBM Directory Server for authentication and security control.
Q: Is the notion of getting IBM's software family integrated a new direction for IBM?It's something we've been working on for a very long time. But it's fair to say that all of these things had their own history associated with them. Until very recently, it wasn't at all clear that the market was looking for this degree of integration.
Q: What's changed?
If you think back over the course of the last 10 or 15 years, people have been building systems that tended to be vertically automated. They were building a credit card processing system, or they were building a customer information system, or they were building an [enterprise resource planning] system. The connection between those pieces was not as important as the features that the vertical application provided.
What's happened over the last couple of years is that people have come to realize that a lot of the productivity gains that were gotten by the vertical integration of applications have now been realized. If they are going to get more gains, it's through a horizontal integration of applications into a set of business processes or business flows that literally go from demand to delivery. This idea of trying to standardize on a middleware infrastructure all of a sudden has become top of mind with companies as they look to understand how they can respond to the economic challenges they're facing.
Q: Where do Web services fit in?I like to tell people that there's nothing that Web services do that couldn't have been done with another technology. It's just that Web services let you do it much faster because the industry has agreed on what the interfaces are going to look like and how they're going to behave.
There have been interface technologies before; there was [Common Object Request Broker Architecture], and Microsoft promoted [Common Object Model and Distributed COM]. But none of these achieved a critical mass in the industry. None of them achieved the de facto standardization that XML and Web services are now achieving.
Web services have become the standardized way with which people can interface their applications. And that's a very powerful, very important technology. But it doesn't change the fact you still have to have robust infrastructure that sits behind it. It doesn't change the fact you still have to have transaction systems. But now it allows me to connect my business with those of my trading partners and my customers more quickly.
Q: Describe the ultimate enterprise integration goal.What all companies are looking for, whether they are in manufacturing, consumer packaged goods, banking or insurance, is connectedness. They're all looking for the ability to have what's going on in one part of the business reflected, more or less in real time, in the rest of their business. They're looking to understand what's going on in their stores, they're looking to make replenishment decisions in real time based on customer buying behaviors, they're looking to have their suppliers hooked up with them in real time such that decisions about replenishment can be made. If you think about the implications of that from a business perspective, you dramatically reduce the amount of wasted inventory, for example. If you make a mistake, you understand the implications of that mistake instantly and you don't perpetuate it. All these things end up leading to a higher overall margin, higher profitability, more efficient business. This is nirvana from a business perspective. Now I don't think that any of us are naive about the fact that businesses are a long way from this.
Q: How far along are we?
I haven't seen any data that would allow me to do anything more than guess. My guess is that we are a fraction of the way there. Maybe 10 percent. We're there enough to know that the benefits are tangible and real.
Q: How is integration technology changing?Now the opportunity is to link the 30 or more years of investment in IT applications to create a view of a new business process, a new set of process flows, if you will, that can be modified and changed by the business as it sees the market conditions change. This notion of business process flows, business process choreography, is going to be a very important notion going forward here. It's going to be as important as the notion of codifying the business rules was in an application environment.
Q: What do you mean by business flows?All businesses have a flow to them, they all have a process to them. Today, many of those flows in fact are not well understood, not well documented, and have large gaps in them where data needs to be transcribed, or transmitted in a non-electronic fashion. And so the process model for the business is highly inefficient.
Q: How can technology help?
Now technology is ready to step in with tools that start to let you take those processes and automate them. It certainly involves connecting the various elements, but it's not just about connecting. It's also about having a model that you're going to measure against. 'This is how my supply-chain process should work, this is how I'm performing against that process.' This notion of feedback and measurement is incredibly important.
Q: I heard IBM is working on vertical-specific business integration tools. Where will these fit in?These business process flows actually end up being patterns that are quite similar industry by industry. They aren't identical, but there's a way that an insurance company, for example, deals with its customers that is repeated across most insurance companies. There's a way that a manufacturing company deals with its channels, its customers, its suppliers, its manufacturing capability that is broadly similar from company to company.
Not only are the flows similar but the applications that people use also tend to be similar, so one tends to find SAP in manufacturing and consumer packaged goods, for example. One tends to find i2 [Technologies] or Manugistics for supply chain and maybe Ariba for purchasing. In other industries, one might find Siebel [Systems] as the front end with maybe a homegrown general-ledger system. What one can start to do is put together offerings that are based on these patterns with the sets of adapters and business process models that will make it easier for people to get started.
Q: What's going on with security?Security has been a big issue for the last couple of years, even before Sept. 11 and of course now after that. We are doing a lot of work with the industry to define what security means, particularly in the context of Web services, which didn't have a security model at all. But we're also working on enhancements to our own directory, and security products for people who are not using Web services. We'll make the Web services and the non-Web services worlds link.
Q: Is some of this available now? What's missing?We have built into WebSphere a set of technologies from our Tivoli [Systems] colleagues. The products, Policy Director and others, allow me to do rules-based authorization. We have a full [Lightweight Directory Access Protocol] directory built into WebSphere, which is where you store name and address and authentication information. So we have built into WebSphere today a full infrastructure for doing authorization and authentication. Now what's needed is for that infrastructure to talk to other infrastructures. So we need to make it talk to all of the other legacy authentication systems in the world, be they mainframe-based or Unix-based or others, and we need to make it talk to the Microsoft-based authentication systems in the world, be they based on Active Directory or Novell's NDS or other things.
Q: How much progress has IBM made on this front?That's something we're working on. We solved the problem in the context of WebSphere, I guess is the message. But the context of WebSphere is not broad enough. We actually have to solve it in the context of the broader community to make it useful for customers who have all of these things. And if they don't have them internally, their suppliers and their customers do.
Q: What do you see as the most promising technologies coming down the road?I'm excited about the possibility of Web services. We're in the very early stages of understanding what it can do. I'm excited about alternative clients, particularly wireless-based devices and how we can extend access to systems away from traditional PCs and to non-traditional devices like PDAs and cell phones. I'm excited about the ability to incorporate voice, as sort of a first-class user interface type, into our systems so that I can speak to my system and have it understand my Canadian accent and provide intelligent responses back.
I'm very excited about how the technology is going to evolve here, particularly at the user interface end. This is where a lot of the innovation is going to happen. The course of the technology as one gets closer to the middleware and infrastructure is somewhat more predictable.
Q: How about grid?
I'm also excited about the possibilities around grids, and the application of grid technologies to commercial computing, particularly to help in simplifying the management of large numbers of heterogeneous computers, which tends to be the norm in many of our large customers today.
It's not unusual to find some of the big banks with thousands of servers, Unix, Intel, mainframes. And all of them are different, all require different administrators, different skills, different interfaces. They can't be administered centrally, as a pool. Grid technology can be of substantial assistance in the administration and operation of these devices. So I'm excited about where that's going to go.
Q: Does WebSphere have a place in grids?Certainly we'll also figure out how to do workload distribution, and WebSphere is a very important part of that play for IBM. WebSphere provides a consistent way of defining workloads, across mainframes and Intel boxes and Unix boxes as well. So one can think of a WebSphere-based grid infrastructure that would give companies the ability to manage and administer workloads and distribute workloads to the most appropriate place in their enterprise.