Getting Schooled on SaaS
A Fairly Understandable Explanation of a Phenom

Dan Carmel is an enthusiastic guy. White-tent evangelist enthusiastic. We were talking—mainly HE was talking!—the other day about the latest emerging trend in, well, just about everything that a business does on a daily basis.

OK; that’s a little overblown, I suppose. We were actually talking about "software as a service," or SaaS as they insist on acronyming it. (I want someone to explain that upper case/lower case thing to me someday.)

Dan is CEO of SpringCM. SpringCM could best be described as a content and document management software vendor, I guess. But there’s a lot of those. Dan’s company has a twist. A big twist. It specializes in delivering its customer applications over the Internet in a hosted environment. Meaning the various application software physically resides in a computer at Dan’s place, while the users can sit at their desks any place in the world.

There’s nothing especially new about hosted software. Many of you reading this have probably used Salesforce.com, or know someone who does. Or probably have at least one work application that is accessed through a Web browser rather than a proprietary user interface on your desktop. So the underlying concepts of "cloud computing" as they call it are familiar and well-entrenched.

But SaaS is something different, as Dan patiently explained to me. We started with a backdrop of the philosophy of SpringCM. "We think of what we do as more of a platform, recognizing that it is a horizontal view of the world. I see it as a whole new genre that I call ‘content-based applications.’ Just like applications in the data world... CRM, ERP, etc. Usually, if you want to automate a process, you take a bunch of technology, call an integrator, and they come out and develop that application for you. The problem is, you get a lot of very, very specific applications. Just from a business perspective alone," Dan adds as an aside, "that’s a tough challenge for the integrator. It’s very difficult to survive on a niche."

So at this stage I naively proposed to Dan that his assessment of the disadvantages of the systems integrator model is exactly what led to the rise of the application service providers (ASP), the ancestors of hosted computing. I struck a nerve.

"The business model around ASPs is totally flawed. There have been business model changes that make true SaaS an absolutely compelling, transformational phenomenon." (Dan says stuff like that a lot; I told you he was enthusiastic.)

"Hosted versus SaaS requires some definition," began Dan. "They share one aspect; they’re both hosted. You get both for a monthly fee, both require no hardware, both have no software installed at your premises; and that’s all good. But that’s where the similarities end."

Dan continued: "ASPs work like this: You give your software license to an application such PeopleSoft to an ASP. The ASP will host it and run it on his infrastructure, BUT—and this is key—for each client, the ASP hosts a special instance of that application.

"So each customer runs its own copy of PeopleSoft, with the ASP’s people maintaining and, more importantly, customizing each customer’s license. It takes the hosting and customization off the customer’s plate, sure. But the ASP has to understand the code well enough to integrate it into the customers’ other applications. They have to get feedback from the customer in order to customize it on an ongoing basis. They need the customer knowledge about the way they work, from the outside looking in, and do all this substantially less expensively than the customer could do it himself. Multiplied across all the many customers they support," he explained.

This, according to Dan, is an untenable business model. "It doesn’t knock enough cost out," he said. "It’s only marginally less expensive to run and maintain 20 copies of PeopleSoft for 20 customers than it is for the customers to do it themselves." So, the one-to-one relationship of hosting, maintaining, customizing and integrating drives the cost through the roof. Plus, the pipes to the customer, during the heyday of the ASP, probably weren’t as fat and reliable as they are today. "Bottom line: it didn’t work," Dan concluded. That "heyday" was probably about year 2000-2002.

Where We Are Now
Move ahead six years or so. According to Dan, "major changes have happened within the IT fabric. One, the browser’s gotten better, so we can have richer applications. Two, the integration paradigm has begun to change. We’ve accepted Web services, for better or worse, and we no longer care (assuming the bandwidth is good) whether that Web service is hosted inside the firewall or outside. That’s BIG. And the pipes are bigger, so bandwidth is larger and more reliable. Final result? We’re now ready to provide a better experience to the enterprise user, and to the IT organization who (at least in terms of the software) can be relatively indifferent as to where the application is coming from."

With all those components in place, explained Dan, the stage is set for a true "revolution" (his word, natch) in business information management and computing. "Into this environment, we can introduce a new architecture to build and run applications: Multi-tenancy." Great, I think. Another arcane technology term to learn and forget within the next six months because it will probably be as tried and worn as an old sock by then. But... maybe not.

"Multi-tenancy means the ability for multiple organizations having very different experiences using the same executable code. In this model, there is only ONE copy of the application... Salesforce.com or Peoplesoft or SpringCM document management... whatever."

I’ll try to sum up the multi-tenancy component of SaaS this way: Unlike the ASP model, where all the customers have their own copies of software—and for which they are paying licensing fees, BTW—in multi-tenancy, the provider (the SaaS provider, specifically...more on that in a minute) has only a SINGLE blessed copy that everyone uses. The advantage for the SaaS vendors is pretty obvious: instead of maintaining PeopleSoft version 4.13 for the Acme Corp. on one machine and version 3.97 for the (slacker) Zenith Corp. on another, and so on and so on, the SaaS vendor offers ONE version—usually the latest—that every customer uses.

Maintenance and upgrades are thus limited to a much simpler, one-touch operation on the part of the SaaS vendor, or as Dan put it: "I’m not dealing with 20 different instances of software. Instead, I push the button on Saturday night (well, I don’t, but someone in my company does!) and by Monday morning, thousands of users are using a whole different version of the software."

Now, I didn’t set out to deliberately look for ways to shoot holes in Dan’s SaaS thesis. I’m actually kind of fond of the idea. But in the interest of journalistic exploration, I had to find ways to shoot holes in Dan’s thesis. That’s just the way I roll.