Tim: Let’s kick of with some examples of problems that are suited to a Web Service solution.
Jason: It’s things like business-to-businesss, supply chain enablement, supply management, inventory, MRO (Maintenance Repair and Operating), where you have lots of different companies dealing with lots of different buyers, all of whom have different kinds of back end systems. These are usually disparate systems that need to be able to adjudicate bills, manage inventory. Web services play that lingua franca role of being able to help communication across these diverse environments.
For example, say you have different organizations within a company, perhaps through merger or acquisition, that have to be able to interact and interoperate. Rather than changing out large infrastructure or legacy back-ends, you’re able to use interoperability standards, web services by and large, to create a degree of communication between systems which natively don’t communicate with one another.
Tim: Can you give me some examples of where they have been implemented successfully?
Jason: We’re doing this all over the world with large customer accounts. Volvo, Shell, these are just the tip of the iceberg.
It used to be if we wanted to build an application it had to be fairly heavy. You had lots of objects, and if you wanted to call the object you had to go through the whole operational framework of an application. Now we’re actually able to build these applications in virtual formats using these interoperability standards and these communication and routing standards.
For us web services is an over-used term. It doesn’t fully encapsulate all of the impact that these standards-based technologies are enabling.
Tim: Some people are questioning whether there is really a high take-up for web services
Jason: We’re seeing a tremendous level of adoption at very senior enterprise levels. For example, we have a couple of projects that we’ll be rolling out here in the next few weeks. One of them I can announce, which is with Pep Boys, a large automotive supply retailer in the US and Canada. They have 700 stores, and they’re looking to grow to a couple of thousand stores. We are building a service-oriented architecture from the ground up with them, which will rely entirely on service-based standards, utilising our mediation technology.
You asked earlier where our applications are going, and WAS [Websphere Application Server] is a great example. WAS is morphing over the next year or so into a complete web-service based environment, where it becomes much more dynamic in terms of routing , mediation, transport, and message management, than we’ve seen before. It is using these technologies to drive all of this. We’re looking for example at semantic brokering in order to accomplish this across domains. So it creates a whole new economy on the performance side.
Tim: What do you mean by “Semantic brokering”?
Jason: If you look for example at the insurance industry, there’s a set of standards, web service standards, called ACCORD. These are semantic standards that the industry has agreed to use relative to the way in which inter-company activity and processes are managed within the insurance industry. The same is true in the Telco industry with a thing called PARLAY, another set of standards.
These are web services, much like BPEL [Business Process Engineering Language], and what they do is to take BPEL to the next level and say, OK, here’s BPEL from a business process standpoint. Now we’re going to attach to it a set of semantic guidelines so this is BPEL specific to insurance, or BPEL specific to finance, or BPEL specific to the Telco industry. It creates a level of specificity that we’ve never had before.
The industries themselves are embracing the standards efforts, and employing them as they evolve their business. So it’s no longer simply a technology issue, it’s now business owners driving the technology. IBM has positioned itself saying, tell us what the business issues are, and we will create enablement protocols to make them happen from a technical standpoint. It’s mapped from our modelling tools, for example Business Integratrion Modeller, and directly form our development tools, such as Rational XDE.
To say that we embrace this technology is an understatement. We have committed to a long term strategy around componentisation and distributed application management and mediation, and all of our products reflect that. The first set of those products comes out in the summer, with the release 6.0 of WAS and the release 8.0 of UDB [DB2 Universal Database for z/OS version 8]. All of the products will reflect this. At some point or other what will happen is the products will be fully componentised, at which point they truly become service based offerings. That’s where we’re moving.
Tim: An early criticism of web services was lack of security, guaranteed delivery, transaction support. How is this changing?
Jason: Forget about XML web services for a moment, because there are other service standards that are not necessarily XML based, such as binary XML. But the big answer to your question is that we have seen in the last 18 to 24 months a significant willingness on the vendors’ parts to agree to agree. We’re seeing this around the WS-I [Web Services Interoperability Organization], which is looking at transactions, mediation, security.
There is a greater degree of agreement now among the vendors because frankly they’re getting to the point where they can begin to support these standards. Some of their earlier servers had difficulty with that. I won’t mention names. Now that these companies are better able to support the standards, they’re more willing to say OK, we agree. We’re seeing a huge acceleration of the standards development around interoperability, transactions and security, and also around management.
At the core of IBMs belief in a mediated environment using web services is that there has to be a very dynamic management infrastructure, one that allows for managing quality of service, that allows for managing service delivery commitment, that allows for managing security and single sign-on issues, that allows for all the contingencies that have to be audited, metered and managed within the context of a distributed, mediated environment. We are deploying this today.
Tim: So does IBM already have the standards it needs?
Jason: We’re far from having everything we need, but … I would say we’re 60% down the road of having the core standards that allow for true interoperability. And they’re hardened, enough for the IBMs and the Microsofts of the world to use. If you look at what Microsoft is doing and what we’re doing, we’re moving forward on the assumptions around certain standards, for example transact and mediation, interop, choreography. And everyone’s agreed around BPEL. We showed with Microsoft last Fall the web service security piece, that we can interoperate and exchange PKIs to Kerberos. We do this now, we can guarantee this to customers, we can show them on an SLA [Service Level Agreement] how they perform. I would say that it’s definitely ready for prime time.
Our recommendation to customers is that they incrementalise. This is not a case of changing the world overnight. This is something that you incrementalise based on business need.
Tim: What about older distributed technologies, such as CORBA and Java RMI Are they being replaced, or do they still have a role?
CORBA is going the way of the Dodo bird, but we’ve extracted from CORBA some of the lessons learned. Some of the code, and some of the fundamental aspects of what CORBA was about is integrated into the new technology. EAI [Enterprise Application Integration] will go away. If you look at the EAI application vendors, they are all retooling their applications now into web services. It’s basically saying, EAI doesn’t work any more and now we have new technologies. To some extent they’ll succeed and to some extent they won’t. These are not companies that are by and large application management companies. From a distributed standpoint they were putting all the spaghetti into a box and as long as it stayed in the box everything was Jim Dandy. But if you want to take it out of the box, it gets messy.
IBM is not in the application business, it is in the integration business. The middle tier is where all of this is going to take place. Quite frankly I think there is only a few companies that can truly stick their stake in the ground and say this is our space. I think these smaller vendors will contribute to some of these niche technologies. I think the standards are hardening and we’re seeing this across the industry. And I think we’re looking at a change in paradigm, a major change in paradigm. It’s the most exciting thing I’ve seen in the history of this industry, in my 30 years in it I’ve never seen anything like this before.
Tim: So how would you summarise the business benefits?
Jason: What are the benefits? They’re outstanding. Loosely coupled applications, loosely coupled services, the use of services for movement of data and application capability, service enablement which allows for virtualization of resources, and preservation of existing legacy systems without the overhead attached to it, the list goes on. The ROI is substantial. We’re seeing huge savings.
If you look at the banking industry where they’re coming out of very large mainframe legacy systems, they don’t have to scrap everything and start all over again. Through the use of adaptor-based technologies and an SOA [Service Oriented Architecture], we can tap straight into those, and have them be service enabled, and use that data through things like CICS or other TPS [Transaction Processing] Systems. They retain all their value. The customer benefits by having a level of management they’ve never had before.
Tim: What about the performance problems with XML?
Jason: For performance XML could be a problem. It depends on how it’s used, and that’s why you want to create a mediation layer. At the mediation layer where the parsing takes place, that’s where we usually do a lot of conversion to binary XML, at which point we’re seeing a 400% improvement in performance. That’s substantial. It’s happening under the covers so the customer doesn’t have to fret about it. They just have to bring the WSDL [Web Service Description Language] or the XML in, and the rest is behind the covers.
Performance-wise, parsing engines are getting better and better. We saw about a 150% improvement when we implemented a new engine for SOAP on our last release of WAS, and we’ll see that continue to improve. We’re also going to see some of the standards around how schemas are composed and how headers are added, and how we manage data, we’re going to see economies, new effective ways of writing the schema to reduce some of the issues on performance. It’s not 100% but it’s 90% of the way there.
Tim: What about versioning? Say you have an existing service and you want to change it?
Jason: We support common service based modelling. The WSDL is nothing more than a service that can be called through the management framework. If you need to change the content within the WSDL, you can do that through the service management console. It propagates down through the system, so everything that was looking for the old service is now seeing it as the same thing. Don’t forget, in an SOA the WSDL is just another virtual service object.
In the framework of a Service Oriented Architecture you anticipate that these things will have to happen and you build a mediation framework that supports it. And you do that through a single console interface.
Tim: Can you say a little more about how WebSphere is evolving?
Jason: WebSphere is our middleware platform. That includes everything from our application server to our business integration technologies to our database systems with DB2, to our management infrastructures under Tivoli, to our workplace management tools under Lotus, and our development tools with Rational. WebSphere is fully integrating all of that.
I came from Microsoft. When I used to look at what IBM was doing it seemed to me a hodgepodge. At a distant it seemed to me to be a bunch of buckets and bolts that didn’t make a lot of sense. When I joined IBM I was given an opportunity to look under the covers. I was surprised to discover that the acquisition and development strategy of IBM software over the last 9 to 10 years has been intentional and focused. What Steve Mills [IBM Head of Software] and his team has done is to build a strategic, dynamic architecture for the middle tier. Most people didn’t see that at first because they didn’t have the overall perspective. Our products are moving towards a common runtime and, more importantly, a completely integrated component-based infrastructure. We’ll see this in the next two years across all our products. They will be fully component based, which means you apply only the components necessary.
That gets to your performance question earlier. You don’t have any of those issues with a component architecture, particularly if you’re using that architecture in a service-based environment, which we believe is absolutely requisite to do. The evolution of that platform is predicated on that set of beliefs.
Copyright Tim Anderson 6th April 2004. All rights reserved.