Software 2007: SAP's new code

SAP co-founder and Chairman of the Supervisory Board Hasso Plattner gave a reprise of his "New Idea" Sapphire  presentation at Software 2007. The presentation covers many concepts (see below), which I wrote about in this post from SAP's Sapphire conference in Atlanta last month, that underlie SAP's new code base, code-named A1S, aimed at small- and medium-sized business and due sometime in 2008.

"It's a new system, with no name yet, and it's not announced, but we are starting to talk about it because we cannot hide it anymore," Plattner said. In fact, at Sapphire there was too much talk about A1S and not enough show. What Plattner is describing in the slide above is in part a page from salesforce.com and other enterprise SaaS pioneers.

Peter Graf, executive vice president of the Product & Technology Group at SAP, told me that A1S  doesn't include all of the concepts outlined by Plattner.  "Hasso's key trends converge into the New Idea, but it is not synonymous with A1S," Graf said. "It's a ten-year journey. Henning [Kagermann, SAP CEO] has to commit to specific capabilities that Hasso, as a 'consultant' advising on architecture and strategy, describes." However, most of the concepts in the slide above, with the exception of in-memory databases, will be part of the A1S suite.

Plattner admitted that SAP was one or two years late in adopting the 'New Idea' for developing a product. "It's hard to deviate from where you are successful," Plattner said. "We simultaneously have to think about the future. Whenever we see that a product starts to really do well, we have to think the next generation. We started a year or two too late with this project. We thought the bubble would go to 2003 but it didn't." He also noted that the stock market doesn't reward companies typically for investing in new generations of products. SAP just announced it would spend $300 to $400 million on new products and SAP's stock halved, Plattner said.  

"We have 3000 people working on [A1S] every day to build the system which is significantly different from what we have but it also shares as much as possible," Plattner said. "This time the system will be locked up--no access to code, but 2500 service interfaces. The scope of the system will be covering most areas...the whole nine yards. The difference in this approach is designing the system to embrace user-centric design as one of core principles. We are at version 7 of UI, and by the time of mass production we will be at 8 or 9."

SAP has finally come over to the on demand side. "The system will be hosted, as Google and salesforce.com have shown that this works. We want to provide the complete enterprise that runs in the cloud. We believe its time now to go in this direction and every year it will become more obvious as a viable alternative to on site computing."  He also advocated HDTV video services as a way to overcome the lack of face-to-face communication. "They will make a huge change when can see with four times the type of pixel quality, see how hands and feet move. You can feel the presence but reduce travel time and nights in hotel and overall cost."

Plattner believes that the SAP community, not unlike open source communities, will be key to the future of software. "There is no question that the future of software design will take place in community," he said. How [software] develops and continues in next 10 years will largely be driven by community," rather than the old way of talking to each customer about their needs. He said that two-minute videos with text and pointer to links on how to solve a problem or create a scenario, as well as success stories, would "spread like wildfire throughout the community." 

Plattner was enthusiastic about creating a harmonized user interface, in which every function is enabled by HTML and browsers and also smart clients and embedded environments, such as integrated with Microsoft Office (Duet) or other office suites, is identical.

He views the New Idea as portal-based and event-driven. "Instead of exposing source code, we expose the models, which can be used by designers, process experts and implementation architects, so they can find the entry points, and which components can be bypassed to build something else, and it is fully documented. If you change process flow in models, they will behave differently. We are driving behavior of the system through models, which drastically reduces the need for table-based customization."  

Plattner was most enthusiastic about in-memory databases. He described taking 36 million line item, 16 gigabytes of data, and compressing it to 1.2 gigabytes in memory. "Any question asked, excluding bringing up a spreadsheet or presentation, any filtering, aggregation or comparison, the answer time is 1.1 seconds, which is directly proportional to the gigabyte size," he said. "One second or millisecond response time, split over multiple computers with 100 gigabytes of memory and 8 to 16 cores--it's a completely diff operation coming." He compared the in-memory breakthrough to Google's breakthrough--returning search result very fast. "If you if don't get the precise answer fast, you can use your brain to narrow it down, rather than formulating complicated queries only to see the response is zero."

 

Plattner views the new code as technology that can influence other SAP products. "We can see what we could do and use it as blueprints for future releases of on onsite computing [products]," he said. "We cannot accelerate beyond a 9 to 12 month cycle, but starting with perfect blueprints is a completely different starting point compared with discussion, PowerPoint slides and half specs and then starting the developerment. It helps accelerate development, and comes back to the development speed in early 90's, when we could roll out a release every 3 months." The 1990's release schedule proved to be unsustainable, but Plattner believes that with the new code the barrier can be overcome.