SAP's Shai Agassi on Oracle's 'folly' and the shift to services

Shai Agassi, president of the Product and Technology Group and a member of the Executive Board at SAP AG, fielded questions from New York Times tech reporter John Markoff and the audience during an early morning Churchill Club event on Wednesday. We have a podcast of the interview, as well as a separate follow-up podcast I did with Agassi, digging deeper into some of SAP's technology, hosted services and his disparagement of Oracle's strategy. I also checked in with an Oracle senior VP Jesper Andersen to get a response to Agassi's comments. The podcast can be delivered directly to your desktop or MP3 player if you're subscribed to our podcasts (See ZDNet’s podcasts: How to tune in), or you can just download the Churchill Club interview MP3 file and my conversation with Agassi.

SAP's Shai Agassi

During the Churchill Club interview, the 37-year-old Agassi offered his reflections on his journey from an Israeli startup to the Executive Board of the German software giant Silicon Valley, an Israeli working for a German company, the socially responsible enterprise, outsourcing, open source, SAP co-founder Hasso Plattner's influence on him and, of course, the competition.

He characterized Oracle as a "big pirate ship" with a deeply flawed strategy. Oracle has an "immature view that we are in a mature industry," Agassi said, making a case that Oracle's aggregation of customers in buying up enterprise software companies on its $20 billion spending binge and the promise of Project Fusion will be disruptive to customers, not to the industry competitors. "Oracle is like landlords in Cupertino [Silicon Valley] in the middle of the bubble," he quipped. Oracle shareholders are being ripped off if Oracle doesn't invest the maintenance revenue stream in innovation.

Fusion will simply be the next version of Oracle's enterprise applications, and Oracle's customer base will be forced to make the big switch, which conveniently leaves the door open for SAP to intercede. Nor does he believe that Oracle's plan to deliver Fusion--the next generation of enterprise applications that take the best features from all the code bases acquired to build a new standards-based ERP platform--in 2008 is realistic:

"There is a fundamental problem in whole notion of Fusion. If you believe you can go out and rewrite from scratch every single process at the backbone of this collection of processes for every industry, and do it in a matter of less than five years and have less than three years of stabilization of code, you are obviously a banker. You really need to think about it in a sense of how much code goes in there--the core of processes for 28 industries...it's the analysis of the genome of every company in the world. To rewrite it from scratch is a suicidal attempt. Maybe they don't write it from scratch, maybe they pick it up from all the applications they bought. The problem is each of the applications uses a different coding mechanisms...if you pick all the legacy languages, you pick all the legacy with it. It's not an approach you can take. We will see an about face and Fusion will be Oracle 13. If all the noise is just to go back to Oracle 13...pirates in the water.

In my conversation, Agassi said that Oracle is basically telling customers they will have to wait until Oracle gets Fusion right, and in the meantime pay more maintenance that before. Nor are Oracle customers incented to buy more software upgrade, given they are looking at a completely new platform--whether from Oracle or SAP--in the next few years, and will simply make do with what they have until they are confident that they can transition without serious downsides.

Nor will Oracle's developer and consultant ecosystem profit much from thousands of customers in maintenance mode, according to Agassi. To top it off, Oracle's database business in less secure with memory-based transaction systems and higher level of abstractions, at the object level, that render Oracle's special database optimization sauce less critical, Agassi said.

But Oracle is not alone in transitioning customers to a new platform. The vast majority of SAP customers are still on the R/3 client/server platform and using its proprietary ABAP programming language, and will eventually have to move to the more open standards-based, services-oriented mySAP and NetWeaver. Agassi told me the 70 percent of the estimated 30,000 SAP customers have extended their contracts for mySAP, but that doesn't translate into deployments. Only 3,000 migrations are expected in 2005. Agassi expects about 5,000 deployments in the next 18 months and more complete customer migrations within five years.

I played Andersen the audio from Agassi's interview (at 16:30 in the podcast) with me characterizing Fusion as a rewrite of enterprise application core components from scratch.

"We are not rewriting Fusion from scratch," Andersen said. "A lot of what we already have is written in Java and uses dynamic HTML, AJAX and [Web] user interfaces. We are not building a brand new set of apps. Many of the areas where overlapping --there are only so many ways to post transactions to a general ledger, for example, and we have been careful how we scoped Fusion."

Andersen did say that Oracle's E-Business Suite (an upgrade--version 12--is in the works) would be the hub of the superset of features that go into Fusion product. Agassi predicted that would be the case, but asserted that Oracle would have problems in assembling a common, modern software framework and set of service-enabled business processes from the various application code bases recently acquired.

Andersen dismissed that idea that the development challenge was too daunting, and said Oracle planned to ship Fusion by the end of 2008. He also cited SAP's slow transition of customers off of its legacy R/3 system.

"Gartner stats show 94 percent of SAP customers are running R/3, over four years after my SAP was introduced. SAP's business process platform [BPP: a service infrastructure and enablement platform that includes NetWeaver] is separate from other [SAP] systems...it's great for ISVs and building composite applications, but the business processes and modules are different that what is in mySAP today. There is no mention that mySAP CRM will move to BPP, and in BPP SAP has one customer model, and mySAP is on separate database."

"Admittedly, Fusion is a big project. The components essential for Fusion--all the technology required for business process management and orchestration, embedded business intelligence--are all part of the Fusion Middleware suite, which exists today. It is the foundation of Fusion, and we are building it across all our product lines to day. We are taking all the great IP, data models, designs, the PeopleSoft user interface and building it into Fusion. We are leveraging the Oracle code, including Fusion Middlware, which uses Java and using open standards. PeopleSoft and J.D. Edwards are certified on Fusion Middleware."

Like NetWeaver, Fusion Middleware includes application server, developer tools, integration services, identity management, portal, collaboration, master data management, Web services management and other core platform elements. Andersen claimed that Oracle has 5,000 Fusion Middleware customers, and thousands of ISVs supporting it. "SAP's only ISVs [for NetWeaver] are big companies, or tiny little companies outside of Waldorf [SAP's German headquarters]. Not a single software vendor that matters is writing applications for NetWeaver. If SAP is successful, that may change over the next few years," Andersen said.

According to a March 2005 Forrester report by John Rymer that I received via Oracle, SAP lags behind Oracle in the application server platform market (Fusion Middlware and NetWeaver). On the other hand, SAP appears to be ahead of Oracle in developing the thousands of services for building composite applications. In April, SAP rolled out 500 services, such as track shipment through a business process, and expects to end up with roughly 10,000 to 20,000 over the next several years. Agassi said that the industry will end up with three or four service collections--SAP, Oracle, Microsoft and...Google? Agassi noted that Google is moving up the stack, by going after data and document stores, collaboration and file systems.

The question is: Will these services, supposedly based on standards, be interoperable across different vendor application platforms?

On the issue of maintenance revenue, Andersen said, "Oracle is only company that for five years didn’t adjust maintenance prices for inflation, and our maintenance contract includes rights to the new version of product [Fusion] without paying any more. Moving from R/3 to mySAP and NetWeaver is a cost." He also noted a few upgrades were in development, such as PeopleSoft Enterprise 9.0.

Bottom line, SAP and Oracle customers are in for major transitions in the next five years, and both companies are building their own standards-based, service-oriented stacks and more proprietary business process platforms. For customers, the situation is going to be a better negotiating position than in past years. I'll be following up to see if I can clear some of the discrepancies arising out of the dueling executives.

During our interview, I asked Agassi when SAP would get into hosted services. His answer was next year, but the specifics were somewhat cryptic. "We don't believe that CRM stands alone. We will provide a different way to consume our CRM offering. It has to integrated with backbone --just doing sales force automation is not helpful."

I also asked Agassi about his World Series of Poker experiences (at 30:36 in my podcast interview with him). It wasn't a topic he relished talking about. "Games are games and business is business....It's a recreational thing I do very, very rarely, when I want to be far away from business." In terms of lesson learned from playing chess and poker, Agassi said, "You've got to plan ahead and there a lot of things that are unexpected...and when it happens, replan." In that vein, he said that he, and SAP, are trying to be consistent, make good decisions and not make the same mistakes twice.

The big bet made a few years ago on completely revamping the platform around services and higher levels of abstraction is a good one, but I don't believe that Agassi and team are really betting that Oracle won't succeed, despite his painting a picture of doom for Ellison and his pirate ship crew....