article: SAP and the future of software (plus commentary)

In an article titled SAP and the Future of Software published today, authors Dawn Kawamoto and Dan Farber of CNET News.com sum up their interview with SAP chief executive Henning Kagermann as follows:

The emergence of on-demand is apparently not the death knell for traditional licensing agreements.

And what did Kagerman say to justify this conclusion?

Consider this exhange (my italics):

Kawamoto/Farber: Nonetheless, all this talk of on-demand seems to be bringing a sea change to the way customers want to pay for their software. They either want to pay it as they use it or at the end. Where do you think licensing models are going?

Kagerman: It will be very stable in the years to come.

We have had this debate for the last four years: Should the licence model be different? Should everything be on-demand?

But I have spoken to many clients and they want to own [the software]. They are happy with this model. So, therefore, there will be some additional new profit models, that is true, but the core will still want to license. You can ask me again in one or two years from now.

OK, interesting. Then the authors try to come in for the kill:

Kawamoto/Farber: Back to on-demand and licensing — some customers complained about spending hundreds of millions of dollars on SAP and feeling not much closer to wrapping up its installation and implementation. Some talked about the lure of an SOA and XML and keeping their investment in SAP, but any future investments will be done with point solutions. What do you say to these types of customers?

Kagerman: I have not met such a customer. But if I met such a customer, I would tell him that over time he will spend more money because it's very difficult to manage such a heterogeneous environment of point solutions — even if it's on top of an open platform.

I would expect that the customer will try to check and test the flexibility of SOAs and bring point solutions in, but, nevertheless, he has the maintenance and managing of different suppliers, etc.

So the heart of the SAP argument is that owning your application software is preferable to using a collection of point-solutions you assemble on some sort of platform (AppExchange comes to mind). SAP will make some net new dollars (euros) off of this on-demand stuff, but it won't impact their core business.

This is the battle line in the traditional software market vs the emerging software-as-a-service (SaaS) market. Are you prepared to run your business on someone else's hardware? And even if you are, are you prepared to live with the limitations this will place on how far you can extend and integrate those solutions with the rest of your back office?

Kagerman, I think consciously, shines a light on the holy war going on right now between the SaaS multi-tenant advocates and those vendors (ASP throwbacks?) who cater to enterprises demanding single-tenant control. In an interview with Computerworld, one of the multi-tenant purists, John Girard, CEO of Clickability, had this to say:

"Hosting their installed apps in a managed server farm does not an SaaS offering make," Girard says, referring to the recent efforts of conventional software vendors. "We throw a party whenever an installed competitor announces a hosted offering. It validates the SaaS model and spells operational disaster for the competitor."

I can see why folks like Girard, Salesforce.com's Marc Benioff and others are insisting on multitenancy. It's the only way they're going to make money on SaaS applications in the long run. Just look at the economics of it - single-tenant solutions mean everyone has their own database, their own server code and custom extensions. And don't forget the custom SLA. The cost benefits associated with the SaaS model evaporate quickly once the provider starts adding up the bill for all of this special treatment.

And the folks who opt for single-tenant solutions know this - except now they've got the worst of both worlds: all of the cost and risk associated with maintaining a customers infrastructure, and none of the scalability associated with multitenancy or traditional licensed software. "But we closed the deal," they'd argue. "Are you in business to execute on a strategy or to close that deal?" I'd argue back, then I'd head for the bar.

Ultimately, there's a limit to the degree of customization that a SaaS provider can permit in a multitenancy setting. This is, I think, what Kagerman is referring to when he says that customers want to own their software. That and they don't want to risk their business on someone else's IT department.

In the meantime, SaaS vendors serving the SMB are gobbling up venture dollars to support the huge infrastructure investments they need to make to run big multi-tenant operations (with three-nine or better SLAs) that are needed to support the business model. Remind you of another time when the VCs were making bets on the come?

Kagerman, like other old-school packaged software providers, can wait for the implosion. With too many providers offering solutions in the same space, SaaS profit margins will ultimately evaporate in a frenzy of price-cutting, and customers will start to experience the outages that come from the resultant cost-cutting. Add in a few disasterous mash-up incompatibilities, and the traditional software licensing model will start to look a lot more attractive.

The gentleman from SAP had it right when he said to ask him again in one or two years from now. Things should be a lot clearer by then. If only we didn't have to live through these next two years to find out.