I'm down in Miami at Software AG's Innovation World. The first innovation was that they changed the name of the annual customer conference from Integration World to Innovation World to better describe the expansion of the stack and the reason customers implement it.
One of the most interesting sessions I attended was an analyst round table. This was a far ranging discussion on how organizations are adopting SOA, why some are failing, and which technologies are likely to be most important in the future, and which are most important in the economic downturn.
One of the topics we discussed was how do you succeed with SOA, and is SOA even relevant going forward. While most of the analysts there felt that SOA was going to be the dominant architecture pattern going forward, it will take the better part of the next decade to get there. Dave Linthicum , always the provocateur, put forth the vision that angry masses were waiting at the gates with to torches, because it's too difficult to implement. Personally, I think a big problem is that the skill sets are not yet there, and that's why we're seeing the failures. Also, the press will soon, or is already starting to experience SOA acronym fatigue. But even if we stop using the acronym, few in the room envisioned a retreat from SOA, because it's really the only architecture that makes sense. However, organizations will need to understand how to incorporate other architectures, including cloud computing and EDA, and how to manage the disparate parts as a whole.
This is an area where I think Software AG may have an advantage moving forward. They are focused on reducing complexity by making things easier to install, and easier to use with integrated Eclipse based tooling. The whole platform is integrated, which Software AG says lowers the total cost of ownership.
Additionally, and very importantly, Software AG is enabling federated ESBs and repositories. The next version of CentraSite Active SOA will have some very strong support for federated SOA. It will even support development life cycle policies, which means companies can ensuring that development processes and policies are monitored and managed. Federated SOA is most likely to be the way most large organizations deploy SOA. It is complex, so any help here is greatly appreciated.
There were some discussions about how this might move to cloud computing, but Software AG is not announcing any plans to support Cloud Computing - at least not until their customers start demanding it.
Another interesting discussion was whether the underlying applications are the backbone of SOA, with BPM on top to integrate processes across the underlying systems, or whether the ESB is the backbone. The problem is that many organizations who have ESBs the center of their SOA universe have not been successful. There was a venture capitalist in the room who reminded us all that money is the center of the universe, not IT infrastructure. This was a useful point of view. In fact it seems that Software AG, a fiscally conservative organization, is indeed following the money. According to CEO Karl-Heinz Streibich, Software AG's software has the lowest TCO in the industry.
One of my favorite Albert Einstein quotes is "Make things as simple as possible, but no simpler." All distributed architectures are complex. There are many underlying platforms. The role of middleware was supposed to smooth out the differences to make it easier, but middleware itself has become complex. Abstracting complexity is the key to SOA success.