RealTime IT News

Software First, SaaS, Services Later

MOUNTAIN VIEW, Calif. -- Microsoft hosted a lecture on the future of software with an emphasis on software as a service (SaaS) . But the host had a different view on the SaaS future than some of its guests.

Microsoft  sponsored the event along with Carnegie Mellon West and the University of California, Berkeley, at its research campus here this week.

The morning lectures featured a whole lot of people with a PhD after their name saying SaaS is here, it's a SaaS world, get ready for it, get used to it. Then came a luncheon keynote by Microsoft's Craig Mundie, who basically said not so fast. Mundie is chief research and strategy officer for Microsoft and one half of the wonder twins replacing Bill Gates as chief of Microsoft's software vision.

Ray Ozzie is the other half, and his speech at MIX07 in Las Vegas yesterday was similar to Mundie's.

"Services is a bit of an overused term," Mundie told the crowd as it lunched. "In this context, when I talk about services, I don't mean the rendering of a human effort for a fee. I mean as the execution of an application in the [Internet] cloud."

Services in the Internet "cloud," meaning the wider Net itself, are only good for low-cycle services, he continued. Not surprising, one of those cycles is delivery of packaged software. Mundie said Microsoft was looking at new ways for customers to "use the network to buy and maintain and use software so the old pains of having to buy and install and maintain it will go away."

Eventually, he said, people will be able to go to Microsoft.com and see a "Start" button just like they have on their desktops now, and he predicted that, over time, customers will choose to mix and match between on-demand services and local installation.

Mundie also spoke of the need to embrace parallel processing and development, which is something only done in high-performance computing circles up till now. "The world of processors is moving in a direction that if you want to go faster, you need to go parallel," he said. "Yet the apps we have will require a new model of programming. Without that model, we will not get the benefits."

He said Excel 2007 does have some multi-core and multithreading support, such as heavy processing tasks on one core and basic tasks on another. However, Word 2007 still doesn't have ideal voice recognition, which is heavily processor intensive, and he doesn't know when a breakthrough will be reached.

Mundie's narrow definition of services and the role of SaaS contrasted many of the speeches earlier in the morning.

Timothy Chou, a former Oracle  vice president who set up the company's on-demand business and wrote The End of Software, pointed out that the nine largest SaaS companies, including WebEx and Salesforce.com , had combined revenues of $1.4 billion in 2006.

"So my conjecture is SaaS has happened, and what we're in the midst of is the next step on this," he told the crowd. The next step is finding critical mass, because the originators of SaaS are mostly business apps, and now innovation is being driven from the ground up, like in Second Life.

John Zysman, a computer science professor at UC Berkeley, echoed this sentiment that services should not mean automation. "Where is the sweet-spot strategy in an organization? Not in automating the routine. Someone can copy you. The advantage is in the imaginative use of tools, reconception of the underlying enterprise, management of knowledge and exploitation of the tacit," he said.

He used General Motors as an example. It can't sell cars these days -- Toyota recently surpassed it as the top auto maker in the world -- but its OnStar service is enormously popular. But he would never advise GM to ditch car making and sell OnStar, because OnStar is a unique service to GM cars, and is a selling point for its cars.