SAP drives enterprisewide collaboration

Executive board member Shai Agassi discusses the xApps platform and SAP's commitment to openness

AS A MEMBER of SAP's executive board, Shai Agassi is a lead proselytizer for the company's commitment to collaborative application solutions. Agassi met with InfoWorld Test Center Director Steve Gillmor, News Editor Mark Jones, Lead Analyst Jon Udell, and Senior Editor Heather Harreld to discuss SAP's xApps architecture and the company's commitment to open platforms.

InfoWorld: There's been a perception that SAP needs to become more open. Explain to us your core value proposition around the xApps architecture and how that's been a response to the increasing openness of these application platforms.

Agassi: What's always been the differentiating [factor] for SAP was integrated software and support of business processes across a large number of customers so that you can amortize your cost of ownership, with openness for customizations. Nobody runs SAP uncustomized out of the box. If you look at the key features of SAP, it's always been open. That was the core of what SAP did: integrated enterprise applications. How come everybody perceives us to be a proprietary, closed, monolithic dinosaur? If you look at Microsoft, I think their chief architect for .Net [said], "Portals are dumb. Portal is a dumb proposition." I mean, [that was] a quote from the chief architect for Microsoft. Here's the openness company. We get it, our customers get it, they don't get it.

InfoWorld: That's a rich client argument that he was making.

Agassi: It's lack of openness. It's a rich client argument in the sense of you don't need to be open, you need to do everything in Office. People are not doing portals today to do what he thinks portals are. It's not just to put weather on a page, it's to do integrated business process across multiple different applications, controlled by the end-user. That's openness.

InfoWorld: Is it becoming more standard to see companies plugging and playing components such as CRM and HR from various vendors?

Agassi: If you look at the vendors that are financially succeeding today vs. the vendors that are having a really hard time today, the successful ones are the ones who built the suites. The ones that are having a really hard time are the best of breed. As bad as it sounds, the financial situation is pushing a lot of the CIOs to go buy from one vendor. Why? Because they don't want to own the issue of "Whose fault is it [if there's a problem]?" It's not whether Siebel has a better CRM than PeopleSoft or us. You can argue that until you're blue in the face.

I talked to the head of a very large news publication and he said, "A year ago I brought in IGS [IBM Global Services] and told them, 'You do everything for us. You do integration, you do all these things.' I ended up the year and I looked at my pocket and there's $35 million missing. Before, my IT stuff was at $50 million -- the whole thing, with hardware [and] everything. Now it's $35 million just to IGS. Great! Integration, fantastic! But I'm out of money. This year, it's $100,000 to IGS."

InfoWorld: Are the integrators ever going to recover?

Agassi: I think that the integrators [go] in waves. If you look back 10 years, a lot of it was built by these IBM consultants who would come in and do CICS projects for you and 3270s were everywhere and you built all these things and you owned them. Then as we went to enterprise applications, it went the other way and you said, "Now I'm going to buy everything and send these guys home." Then as you start bringing [in] the best of breed, interestingly enough somebody [is] needed to weave it. You get consulting from PricewaterhouseCoopers to buy this piece from there and that piece from there and this piece from there, and guess what? You need to integrate it somehow. Who do you bring [in]? PwC. [Now] it's swung the other way around. Now people are saying, "Well I can't pay these guys so much money." So it's swinging the other way again.

InfoWorld: Because of the economic situation, there has to be an incremental ROI. How does SAP manage that transition?

Agassi: If you look at what xApps are, they're smaller processes. It's not, "I'm coming in with CRM." I'm coming in to solve a smaller, bite-size chunk of your problem. Plant management is a very small problem -- shop floor management vs. the overall running of a nuclear plant. When you come and say, "I don't need to replace your underlying applications, I don't need to change the infrastructure for you. [I] can come in and do it in three months," that's what people want to see. Three months, $300,000. Put in an xApp, don't replace my pieces. Now let's do the next one. And they want to do it one after the other, after the other, after the other.

You don't need to go buy See Beyond and somebody for workflow and Documentum and Business Objects and eRoom and put the whole thing [together] and figure out, "How do I integrate my integration?" That's where you are today. Look at these projects. You buy a Plumtree and then you figure out, "How do I integrate my Business Objects into a Plumtree?" God, it's stupid! This is supposed to integrate your stuff, and you're trying to integrate the integrators. Then you're going to build applications and you're going to need to integrate the integrated applications. Instead of solving the problem, you're building a bigger problem.

InfoWorld: How do Web services play into this?

Agassi: In your [Web services conference] panel, there was [ CEO] Marc Benioff on one end who said, "Web services are services supplied through a Web." And on the other end, you had these guys who said, "Web services are about these yet-to-be-defined protocols" -- it has nothing to do with [ASPs], it's just the protocols: SOAP, WSDL, and whatever. You know what? Everybody's right because nobody knows what it is. Nobody knows what it is because it's in the middle of a sea change. It could go one way, it could go the other way. The definitions are forming right now. The only thing we agree on is that there is a set of standards that are being formed.

We're saying these standards are going to emerge. What do they do to the industry? They don't change radically the fact that you want the single throat to choke as a CIO of a large company. Fundamentally, I want to know: What do you do when your ASP dies and your data goes with them? How do you see beyond the See Beyond? What do you do in those cases? It's a tough question. What do you do if Marc Benioff tomorrow says, "I ran out of money." What happens to your sales force automation software? Where's the managed service value?

We're not saying it's a bad model. You'll see Web services allowing us to build more and more of these xApps faster. It'd be easier if Rick [Bergquist, CTO of PeopleSoft] came on stage and said, "All the interfaces are Web services." I've been trying to do an exchange of software for the last two years. I have an e-mail from Craig [Conway, PeopleSoft's CEO] saying, "We will never give you our software." And here we are talking about openness. Somebody's blaming us for not being open.

InfoWorld: Do you regard workflow and business process integration as something that's not likely to be standardized because your value-add is beyond that?

Agassi: No, I think you'll see some things. BPML is great. BPM, the initiative around business process modeling -- they're all great initiatives. But what you'll see is that there will be agreement on the base foundation. We all need to define the business process; the model needs to be defined, it needs to be read, it needs to be written -- we all agree upon it. But then you're going to get to the minutiae on the detail and the nuances, and that's where agreement is going to stop. Look at what happens in my industry where I've been through portals. We all agree that portals need to use pieces that come in HTML and that they shouldn't bleed from one to the next. So if I put a portal component, it shouldn't bleed to the next component. That was all agreed upon. Once you go beyond that, nobody can agree on how you do eventing. But it's crucial for building complex applications.

Look at JSR [Java Specification Request] 168 in our industry. What has been agreed upon is really the lowest common denominator. Why? Because even Microsoft and IBM, which steamroll these committees, can't agree on anything beyond JSR-168. Great, JSR-168 is part of the application platform server. But all the smarts in portals go into the smart stack. Same thing for workflow. JMS is agreed upon. We know that we need to send a message, it needs to be in XML, it needs to be read and written. We can't agree on anything other than that yet. I think that we will be able to agree to the fact that there needs to be a repository of business models [and] it needs to be read and written. We won't be able to agree as to what's in the repository, but we'll need to be able to agree that there is some way to send the model and read the model.

InfoWorld: Do you think what Microsoft's doing with Yukon is going to accelerate what you just talked about?

Agassi: No, I think Yukon is a question of how you're implementing it. Microsoft is really in the business of stores. They try and own more and more and more of the stores. It's a great business. I wish I was in that business. But that's their business model. We are in a model that is a bit different. We realize that you won't have one store for anything. I mean, it's almost impossible to think that you will have one store for documents across your whole enterprise. By the time you've fixed everything and it's all in one store for text and all in one store for data and all in one store for your customers, your CEO [will] say, "I'm going to buy this company right across the road." And they have a different store and you're back to the same spot.

If you look at the Web services stack, one of the key services is master data management. It's a service that allows me to do very quick aggregation of data, very quick cleansing of the data, and ongoing maintenance of the master data, so that if I change something in one application, [if] I add a customer, I know how to propagate it. Not just in the sense of sending the message, but [in knowing] what is the application logic that needs to go in there. Then I can give you an interface where you can come in and not add my customer into Siebel, but add it to the master data server. And it will go into Siebel and to SAP at the same time. That's the kind of thing we're seeing. By the way, "Who owns the customer at that point?" is a dumb question. We don't think anybody owns the customer, just like nobody owns the employee because they run an HR application. At the end of the day, the customer is the customer [and] nobody owns them. If they go somewhere else, that's your problem. A CRM application does not own the customer, it just stores data temporarily for you until you decide it doesn't do the job for you.

InfoWorld: What elements of the stack do you keep to yourself?

Agassi: Whatever there's no agreement [about]. We're in all the [standards] bodies, we're in all the committees. The thing that we don't do is we don't take over the body. Coming in and arguing against IBM and Microsoft and these committees is not going to make us successful. Coming in and participating and trying to show the angle of what you do with these standards, that's what we do in these committees. We don't try and force our thought process through the committee. We try and participate in it and understand where it's going. And what we'll see is that some things will migrate from the top stack, the smart stack, down into the JSRs.

InfoWorld: Other than the business process space what else do you see migrating to the smart stack?

Agassi: We see 15 services that are going to be key. Personalization, visualization -- visualization is a very underserved market. One of the biggest problems that you see in applications is that every screen is different. So the poor end-user needs to learn my way, your way; even within our applications they need to learn 15 different ways. What we found out is that if you put templates, presentation patterns, and you run all your applications through these patterns, you get usability-to-go factors higher than you had before. Customers are coming in and saying, "This is what I want. I want to learn three patterns and that's it." Even though I don't get the flexibility and everything, I get 5 percent of those screens to be completely flexible [and] 95 percent of the screens come into the three patterns. So you can visualize the same data through the same patterns even though it's coming from somewhere else -- your legacy applications, wherever it is. But the end-users now feel like it's the same screen.

1 2 Page 1
Page 1 of 2