SOA "Soft Landing"
SOA should be no exception. Where is SOA on Gartner's Hype Cycle? Phil Wainewright sees SOA at rock bottom. Tim Bray is already eulogizing SOA and talking about the emergence of something he calls Web Style. Loek Bakker says that SOA has been hijacked in a recent blog posting. Todd Biske calls SOA Battered and Bruised.
Interestingly enough, amidst all of the doom and gloom are some efforts to "reinvent SOA". Obviously it's in the best interest of some companies to reinvent SOA in their own image. Of course one of the great and terrible aspects of SOA is that it operates at an architectural level of abstraction. This allows for a lot of maneuvarability with respect to definitions.
Oracle, for example is already working on what they call SOA 2.0, an idea which they touted at the recent JavaOne conference in San Francisco. "SOA 2.0 is the term that we're using to talk about the combination of service-oriented architecture and event-driven architecture," said Steve Harris, vice president of Oracle Fusion middleware. This term was also picked up by Yefim Natis, distinguished analyst at Gartner who said that the difference between "SOA 2.0" and good old fashioned SOA is the inclusion of "Event-Driven Architecture".
Some of the tension can be found at the interface between lower level SOA concerns such as transactional SOA systems and systems focused on ERP and the more coarsely grained "agile" business service layer on top. Do we need a new name for SOA in this flexible new incarnation? Implementation strategies also represent the types of schisms that exist in many implementations. Document style or RPC style? REST or SOAP? AJAX or Smart Client? ESB or WS Fabric? Data centric or Process centric? EAI or Mashup? Centralized governance or distributed? Taxonomy or Folksonomy (tagging)? Schemas or Microformats? It's a long distance between COBOL based mainframe transactional applications and green screen scraping and social software, blogs and wikis.
Some of the naysayers are no doubt experiencing implementation pain. It turns out that while SOA pilots are easy and incremental, serious enterprise scale SOA requires planning, governance, organizational change, cultural change and real budgets.
SOA in total is also experiencing some degree of schism. While the blogorati and thought leadership are bemoaning the fate of SOA, industry watchers like Joe McKendrick suggest SOA may be on the Gartner slope of enlightenment. Real World SOA case studies are emerging. Organizations are realizing the connection between SOA maturity, SOA governance and business value.
The most significant data point is in the recent study by Merrill Lynch analyst Kash Rangan. The study found that of 76 CIO level executives surveyed, almost half expected IT spending to rise in 2006. Among these CIOs, Nearly all of them -- 87% -- said they believed developing a service-oriented architecture (SOA) to be the "next big thing" in enterprise software. So a very hopeful sign can be found simply by "following the money".
Where's the competition?
I think this is an interesting statement. But take away SOA and what are enterprises going to need to do to integrate their business proceses/systems. How do the ERP, HRMS, B2B, SCMS, etc. going to work together? This will require planning, governance, organizational change, cultural change and real budgets. SOA is just a solution, when the problem changes or more viable solutions are introduced then I'll worry.
No silver bullets, big surprise
The basic problem is that for five years, we've seen desperate enterprise software vendors take a simple set of architectural principles and try to market them as silver bullets. As you note, folks are now finally figuring out that at the end of the day, when it comes to getting complex systems up and running, there's no substitute for old-fashioned elbow grease.
I think that's where the naysaying comes from. It's not directed at SOA per se, but rather at the vendors' SOA 'products' that were supposed to make everything easy. The basic SOA principles are always going to be useful: break your app into functional chunks that expose well-defined, coarse-grained contracts and talk to each other with XML. Those are pretty old ideas for the most part, and I don't think they're going anywhere.
The vendor-fueled hype, however, is hopefully dying a well-deserved death.
CIOs vs Bloggers
There's no bubble that is going to burst because SOA has materialised on the ground. I was there when CORBA was talked up as the choice for distributed integration. 10 years of CORBA talking, teaching and preaching never led to the kind of actual adoption that SOA already has. The need for distributed integration clearly exists and there are no credible alternatives to WS-* at this time when it comes to integrating systems that need security, transactions and support for complex business processes.
Also, I wonder when the bubble bubble bursts. Since the dot com bubble burst in 2000, we keep looking for tech bubbles and we forget that not every hype or fashion is a bubble in the same economic sense as the dot com bubble. Let's look at the economic fundamentals. A bubble is when huge amounts of money feed the supply side and it later turns out that there is not enough demand to ever recoup that money. Is that the case with SOA? I think not. There's no sign of that whatsoever.