Why Software-As-A-Service Could Be A Dead End

With this summer's release of Blizzard Entertainment's Diablo III--and my sons' excitement over the blockbuster game--I gained some insight into where the software world is headed. Think that conventional software licensing, whose premise is "you don't own the bits; you own an entitlement to use the bits", reduces your business agility? That's nothing compared to the can of whoop-de-do that's been opened with software-as-a-service.

The Diablo III "innovation" that caught my eye once my sons started hacking away at it was this: Although there's a DVD and PC installation, there isn't any way to play offline. I found this out because the Internet in my house cuts out at midnight. It's my sons' ex-infrastructure-guy father's way of telling them to give the online world a rest.

In the past, with games like StarCraft, once the Internet cut out, my sons could still play the single-player game. Not so with Diablo III. As young men do, my sons researched whether there's a way to play offline, and one of them reported back to me: "Dad, code for their game is running on the servers. You don't have the whole thing on your PC."

[ Would your enterprise do well to take a page from Google's book? Read Why MBAs Hate Google—And You Shouldn't. ]

Whether it's a bid for stronger digital rights management or more control over in-game cheating, this "hybrid SaaS" approach got me thinking.

Most old games have an incredible lifespan. My sons are still playing, through the magic of platform emulators, some video games I bought 20 years ago. Some of the games are from companies that aren't even in business anymore, and the only way to legitimately get them, if you don't already own them, is to find used versions. But if their producers had used Blizzard's method, nobody could ever play those games again.

That's regrettable in a small way when you're talking about video games, but it's disturbing when you're talking about business software. After all, some enterprise applications have massively long lifespans. There's a point-of-sale system that I wrote well over 20 years ago that's still in use, to the benefit of (admittedly few) customers who are still running it. I spoke just the other day to a colleague who also has a system he wrote decades ago that just won't die. This isn't a unique situation.

If enterprise apps all become SaaS, that changes things. Your software investment will have a very precise lifespan and return. You won't have options or flexibility in terms of running software with in-house, third-party support (for example, the support available from the likes of Rimini Street, which offers less-expensive ERP support for SAP, Oracle, PeopleSoft, and other enterprise applications), or buying Jonathan or Trevor a beer to run a reindex on the system that one of them wrote 20 years ago.

Yes, software-as-a-service provides unprecedented flexibility, in the same way that renting office space does. But like rented office space, it comes with a loss of control and an inflexible, single-party financial commitment.

Global CIOs: A Site Just For You Visit InformationWeek's Global CIO -- our online community and information resource for CIOs operating in the global economy.

My prediction? Someone, somewhere will get burned on that loss of control, and there will be at least a partial swing back to on-premises code for critical long-term systems.

Last week at the InformationWeek 500 Conference, SAP co-CEO Jim Hagemann-Snabe predicted that the future of SaaS will include the option of managing systems on premises or ceding control to the cloud. He made a case that core strategic business--the business you don't want your competition seeing--will likely remain company-controlled, and "edge" cases will be "outsourced to the cloud."

Though he's a biased purveyor--most of SAP's software is on premises now, though its acquisitions of SuccessFactors and Ariba are promising--I think he's right. Not all enterprises rent all of their office space, and the future of software won't be all-rental either.

Couldn't make it to the InformationWeek 500 Conference? Join us for the InformationWeek 500 Virtual Event, featuring the best of the conference plus all-new material. It happens Oct. 2.