Service myths

One of the great challenges of moving to a service-oriented architecture is adopting a service mindset. A few years ago, traveling the conference circuit in the days when application service providers were a hot new property and I was one of the movement's leading lights as the founder of the ASPnews.com website, I used to warn that migrating software to the ASP model "isn't just a relocation exercise," meaning that it isn't enough simply to transfer on-premises software into an ASP's data center without adapting it to its new role. The same is true today of services; you can't just put a WSDL wrapper on an existing resource and expect to get all the benefits of a service-oriented architecture.

It just so happens that the ASP model is now starting to come back into vogue — although nowadays people prefer to call it 'on-demand' or 'software as services' — and this has given me the opportunity to return to writing about the topic as a contributing analyst with industry research firm Summit Strategies. I'm still going to continue spending the majority of my time working on Loosely Coupled, but I'm now also going to be researching and writing reports for Summit every couple of months or so. My guess is that the two topics will converge over the next couple of years anyway, so it's not really a detour, more a pre-empting of the near future.

My first Summit report was published last week: From Myth to Reality: Traditional ISVs' Evolving Software-as-Services Strategies. It details the learning process that ISVs including SAP, Oracle and Microsoft have gone through as they've come to grips with delivering software in a services model. Most of them started out with a set of false assumptions — the 'myths' of the title — about how the ASP model was going to work. This report deals with how they've managed to overcome some of those myths, and what they've started to learn about those changes they have to make in order to succeed in delivering software as a service. But although they've made a fair bit of progress, they still have a long way to go before I'll be satisfied that they've really understood the ASP model, and that's the subject of my next report, which digs more deeply into the characteristics of the software-as-services model, and the true nature of the threat it poses to the conventional software industry. More on that later.

What prompted me to mention this here was an item Jon Udell posted to his weblog a couple of weeks ago, Next-generation infoware. One of the myths about ASPs has always been that they'll fail because people won't want to entrust their data to a third party. This has always been an absurd myth — by the same logic, businesses should keep all their cash on-site rather than having banks manage it, which of course would be ridiculous. But the focus on data has always been missing the point anyway. It's not the data itself, it's what you do with it that matters. Process is the thing that businesses don't want to have third parties in control of. And the irony of course is that traditional software is suffering a backlash precisely because it forces companies to yield up control of their process automation to software vendors and their systems integrator collaborators.

What Jon pointed out was that the latest generation of online web services providers are leaving users in control of both data and process. We're talking about software providers that don't even need you to give you their data. They simply add process to it by interacting with it, and if users decide to discontinue those processes, they simply withdraw their interaction.

This a great example of how far out of the box people are going to have to think to really take advantage of service-oriented architectures. As Jon points out, even a leading light of the online services revolution like Amazon hasn't fully got it, because it still tries to own user reviews rather than simply linking to them in some kind of value-added aggregation or syndication model.

Greg Gianforte, the CEO of CRM provider RightNow Technologies, likes to say that we're just at the beginning of several decades of exploitation of the software services model. Jon Udell's examples of next-generation infoware are a great illustration of just how far we still have to travel.

posted by Phil Wainewright 10:29 AM (GMT) | comments | link