Back in 2002, recognizing that on-demand applications couldn’t make progress without the emerging technology of web services, I switched my attentions from SaaS to SOA. Five years on, the tables are turned. I’m back writing about on-demand applications, and SOA and web services are very much at the heart of SaaS’s new-found success. SOA itself, however, is facing difficult questions — many of them recorded in my ZDNet colleague Joe McKendrick’s SOA blog. Could it be SaaS’s turn to rescue SOA?

The other week I was in Dublin and I had the chance to catch up with Annrai O’Toole, CEO of Cape Clear Software, one of the pioneering vendors in the SOA space. I was delighted to learn that O’Toole is a complete SaaS convert, on many levels. Firstly, as an avid user of online applications including Gmail and the superb Mozy online backup. The entire company, he told me, is adopting online solutions like these rather than running its own servers. But much more than that, O’Toole is betting that Cape Clear’s future lies in serving the SaaS market, even changing to a subscription license model last year, because of fundamental shifts in the way enterprises consume IT. “The days when enterprises want to host all their own infrastructure are finished,” he told me.

Coming from the CEO of a business that started out selling into the enterprise infrastructure market, that’s a pretty strong statement. Cape Clear was one of the first companies to come to market with a web services integration server — a category that subsequently become known as enterprise service bus (ESB), seen by many as the core component set of an SOA infrastructure. Through my SOA website, Loosely Coupled, I tracked the evolution of that market from 2002 onwards, connecting with O’Toole and his team every couple of months. The last time I met O’Toole was late in 2005, when he was telling me that the next big thing in SOA was the introduction of two new standards called Service Component Architecture (SCA) and Service Data Objects (SDO).

To be honest, I never got around to finding out about SCA/SDO. At the time I was focusing more on this blog, and I guess my failure to investigate SCA/SDO was one of the reasons I gradually left off posting to Loosely Coupled, which has remained dormant ever since (though I constantly hope to revive it). Anyhow, it turns out that losing interest in SOA at that time was something of a lucky break. O’Toole himself soon after began to lose faith in what he calls ‘Big SOA’. All these standards — SCA/SDO amongst them — are more about helping the vendors justify expensive tools and platforms than about achieving results for customers. O’Toole complains that it got so bad that Cape Clear was seeing customers send out RFPs that went on for more than a hundred pages, yet without any clear articulation of what they actually wanted to achieve out of the project.

So Cape Clear has instead switched its focus to working with on-demand vendors (another ZDNet blogger, Dana Gardner, has just published a podcast interview with O’Toole on this topic). ERP startup Workday is its marquee SaaS customer, but it has others, both in the B2B and B2C markets. One of the things that enthuses O’Toole about working with such customers is that instead of simply trying to tick boxes for standards compliance, they’re actually adopting Cape Clear’s ESB technology to solve live business problems. One of the things that ESB has evolved to encompass is support for composition and execution of long-running processes that often depend on external events before they can be completed. In many SOA environments this is seen as a future milestone to work towards. For Cape Clear’s most recent customer wins, it’s a live business requirement. Whereas SOA seems to have lost its way in the enterprise IT market, on-demand vendors are giving it a new-found sense of purpose.

进一步阅读请点击>>>