[service-orientated-architecture] LinkedIn "SOA is Dead Debate"

[Gervas Douglas]

Discussion at http://www.linkedin.com/groupAnswers?viewQuestionAndAnswers=&gid=36604&discussionID=3807474&commentID=4123986&trk=NUS_DIG_DISC_Q_ucg_mr&goback=.hom#commentID_4123986 Who's getting sick of the SOA is Dead blog posts? Here is a reference to another one but this one (the one quoted) makes some good points about technology adoption. http://it.toolbox.com/blogs/the-soa-blog/soa-sponsorship-and-the-business-31915 Posted 9 days ago | Reply Privately Eric Roch Chief Technologist at Perficient See all Eric’s discussions » Comments (9) Luis Espinal Software Engineer Sr Staff at General Dynamics C4 Systems Absolutely. It sounds like sour apples to me. To be honest, the concept of SOA has taken a hit, in part by the current economy and by IT shops that have failed miserably to implement one (they can't even web sites right anyways.) But the concept is not death. It's evolving. At the end of the day, the bloatware that is attributed to SOA due to the WS-* stack is a problem with the over utilization of WS-*, not on SOA. SOA is not WS-* nor it needs it to be what it is. If one were to say that SOA is death because it's bloated (due to WS-* or whatever) and because most companies that have tried (or claimed to have) have failed (even if they think they have not), then we could also argue that relational database theory or object-orientation have failed, too. A large number of shops have failed miserably to implement a good relational database, instead having a set of inconsistent flat tables on top of a RDBMS. And one is just to take a look at the code running most "OO" systems (specially in Java and .NET) - procedural hyper spaghetti cross-cutting plain old, flat data structures dressed up as classes. Hell, we could also extend that argument and say that software engineering as a whole has failed. It is the individuals, the shops, those who claim to be engineers that have failed to scope things out correctly and implement things the way engineers are supposed to. It is a personal failure, not an intrinsic, fundamental failure of the technology. SOA is going to evolve and be different from what people thought (five years ago) what it would had to be. If that's death, then evolution and adaptation are synonyms of death as well. What we understood for good object-orientation has changed over the years. What we understood for good software engineering principles have also changed. What we understood for software architecture has equally changed. The principles of partitioning capabilities into services, of management, of IT governance, of registries, they are all present in one architecture/management paradigm or another. Calling X death because we have Y is putting lipstick on a pig assuming that Y is the perfect and eternal solution that will never be subject to evolutionary stress and without considering that X might (and most likely still have) value. I'm currently working on a successful SOA project. Whether it will remain successful for as long as the Big Guys think it will, that's something else. But then again, that's the same force that acts in all software endeavours. It is hardly a reason to say "The King is Dead, Long Live the King".... .... unless we want to market-viralize whatever techno kool aid that happens to be the new chic thing to nut ride. Posted 9 days ago | Reply Privately Steve Horsfield Experienced software architect/developer (SOA, .NET, ASP.NET, WCF, WF, C++, Java, JEE, SQL Server, Oracle DBMS, BizTalk) Luis - nice comment - a little aggressive, but fundamentally true. Over the years two main things have changed: 1) Our understanding of what an ideal system should look like 2) The availability of resources (people, time, off-the-shelf code, processing capabilities, etc.) But how often have we heard those dismissing SOA fail to recognise why SOA came into being in the first place and the fact that many of their "SOA" projects didn't fail because they were SOA - they just failed like so many other IT projects. Technology doesn't define SOA, although having agreed standards does improve the potential for adoption. Personally, I think lessons are still being learnt but SOA has at least brought to the mainstream a recognition of some important good architecture practices, most notably the need (or at least consideration) for loose-coupling at all levels of a system. It is very similar to database design - you consider loose coupling, just like table normalisation, and then you optimise where you need to. And before I get those responses, yes - I know SOA isn't defined by loose coupling!!! Nevertheless, I see it as one of the most important and fundamental shifts in thinking that came with the SOA era. To me, that marks SOA as a success. (Regarding projects, I'm also working on an SOA project. The SOA side has delivered everything it promised. There are issues but they are project-level and having nothing to do with the architectural paradigm). Posted 9 days ago | Reply Privately Kevin Puscas IT Architecture Solutions and Consulting I think part of this is just the classic "trough of disillusionment" that all over-hyped technologies go through. And like most the next phase will be seeing SOA as just the way of doing IT. I do think one of the things that has led to the trough (aside from the usual tech media and vendor hype) is also one of the most important outcomes. Mainly bringing the business side of things into the solution discussion. In my experience with various clients the business was simply not ready for that discussion. The whole idea of IT portfolio management is still foreign to most organizations. Most business see themselves as the customer of IT like I'm the customer of a guy putting a new roof on my house. I just want the results as cheaply and as fast as practical. Ok maybe not the best analogy but you get the idea. As long as business maintains this relative hands-off attitude the overall promise of SOA, beyond the technical benefits, will be difficult to realize. Posted 6 days ago | Reply Privately Vasudevan Mooss Modernization Consultant Any SOA initiative which is not connected or driven by the business is born dead. Nothing can save it. Most of the time enterprises seek the help of external consultants and service providers only after they (IT dept) have tried to go the SOA way multiple times and have failed miserably. I have met with some Enterprise Architects who fear to spell the term in any meetings or sessions with the business fearing a backlash. Business folks feel that IT community has unleashed the term only to keep them on the rolls. Posted 6 days ago | Reply Privately Kwong Chi Lee BEng(Hons) Applied Computing and Electronic, MSc Advanced Computing For me I would think SOA is not dead at all. Although I am currently not working on any SOA project. Whenever I see a system, I would think SOA is the most suitable solution (There are many case studies in IBM, Oracle, Microsoft, and etc... to proof SOA is the solution). When a new concept is introduced, a lot of learning is required and if not applying correct methods to the concept, the concept will failed. The SOA concept does give the quickest response to change IT systems to meet business need. If using a tightly coupled system and if business processes are changed, then does it mean the system need to be re-writeen or re-design? But for loose-coupled system with correct tools, the system can be changed dynamically, maybe without the need of highly skilled developer to change the system. I remember one of the SOA promises is also to provide a tool for non-IT/engineering people to manage a system to reduce all the handling costs and lose of information during handling. In my master project, I used netbeans as the development tools and to add a "Send Email" activity to a business process, I simple linked the attributes I required to the "Send Email" web service, re-build and re-deploy the application (It was only a 5 mins job). But for tightly coupled system, how long it will takes to includes this, and at what level the system need to be modified? For me, simple answers to the questions above make me truely trust SOA is not dead and is on-going. Posted 5 days ago | Reply Privately Clive Jefferies Research student at Keele University Interesting thread. Posted 5 days ago | Reply Privately Kathleen A Keating Principal, Public Relations Consultancy It's time to let the dead SOA story die and get back to work. http://www.weblayers.com/SOAblog/ Posted 4 days ago | Reply Privately Luis Espinal Software Engineer Sr Staff at General Dynamics C4 Systems Oh c'mon. There is nothing like beating a dead horse to a pulp just for fun and giggles! Posted 2 days ago | Reply Privately Michael Poulin Proof of Solution Manager at PADA "It's time to let the dead SOA story die and get back to work" and remember the lesson: think in SO, architect and design in SO and watch after implementation to prevent it doing RPC and OO across services. Posted 2 days ago | Reply Privately Gervas