EDABlog

There is a huge amount of hype around SOA especially around the ESB.  Companies are wasting a fortune and it has been going on for a long time now.

Here is one from 2005
http://blogs.msdn.com/richardt/archive/2005/05/04/414833.aspx

Richard tried to distinguish from the hype by using the term just Service Oriented (SO) instead of SOA.

The end result is companies are wasting HUGE amounts of money on consulting repositries frameworks etc . The thing with Service Oriented is its light and it doesnt matter ALL what matters is converting or wrapping your legacy code and that has been happening at a poor rate - mainly due to the money wasted on new Frameworks  , Consultants and high end products.

Do you consulting ?

NO - unless you have no experience with distributed systems or are unwilling to higher someone with such experience.

Do you need a repository ?

NO - some peaces of paper on the wall with a URL and what they do will achieve the same (discovery and reuse) . Its nice but NOT needed. Web apps have the same issues and have done without repositories and management.

Do you need a vendors framework ?

NO - the whole beauty of SO is interoperability and when using something like Windows Communication Foundation its NOT that hard. You do however need Windows Communication Foundation .

Do you need an ESB ?

NO- in most cases these are simple message routers or a PUB /SUB eventing system with a repository management system. If you need eventing there are cheap alternatives ( eg Roman Kiss has a WS_Eventing system on codeproject its simple and works) .

What about security ?

Only if you need it - external communications, B2B etc . If the system your replacing sends information unsecured its not needed for the new system . Convert the systems , introduce nice to haves later - security can be expensive.


Companies goals should be very simple , ie write services to be services oriented to increase reuse. Nothing more nothing less. Its a simple argument SOA services are cross platform and reusable.  Success will depend on the critical mass of existing applications being converted not whether some new minor apps are service oriented. Spend your money on this create a simple app and copy it. Config file deployment is sufficient.

Often companies IT budgets are very limited a significant chunk goes on maintaining legacy system and infrastructure. Whats left is for new developments SOA is often part of this budget but once you start a heavy top down SOA this budget is gone before you reach critical mass.

Once companies have a critical mass of systems being Services Oriented than they can consider some of the enhancements.  Especially since Dublin will have a lot of nice features.