I have been a bit frustrated lately. We are in the process of trying to incorporate a Service Oriented Architecture at work. This is something that I am all in favor of. The upside to this would be the ability to incorporate 30+ years of development and learning. That is something that has an inestimable value to the university. CICS has many tools that can be used to provide SOAP services to clients outside of CICS making all that code usable from anywhere. That's the simple part.

The hard part is that in order to do this in a meaningful way there is still a lot of hard work that needs to be done. Outside of learning the base level of coding it there is a shift in thinking that needs to happen. Implementing an SOA will, if done right, involve abstracting the data that is communicated to applications away from the databases that they live in. That is a hard leap for many to make after years of having the data and presentation tightly coupled.

Additionally part of our current process is the attempt to generate a reusable process for creating,porting and updating applications. This is what is causing some heartache. In management terms there is a need for some controls so that development and work can be measured and monitored. Those tools will be invaluable to the organization in terms of how we plan projects and follow them through to completion. Currently our plan is to work through several projects integrating web services into the mix to open up legacy CICS programs to a spiffy new web front end application. Which in a nutshell is the shiny prize promised on the SOA box. However when you turn the box over and look at the fine print you see the warning. That warning reads something like: You will need to decouple you application from the data store that is the lifeblood of your organization. This really isn't something you can skip.

Right now it seems like we are skipping it. We are basically opening up the CICS side of things and mirroring the functionality in a one to one fashion. CICS transaction NNNN is now going to be a web page at URL. That's nice for a quick win and all but there is nothing reusable there.

I have tried, as well as others here, to explain the importance of coming up with some common representations of the data structures we use. Unfortunately we haven't been able to successfully get the light's to come on. The concept seems clear to most involved. The practical side however, not so much.

I have lately been resigned to the fact that we will be continuing down the current path until the necessity of common data structures presents itself in a painful and explosive way. Sadly, I think there will wind up being casualties along the way.

I have been telling myself each morning that today is going to be a good day, people will see what needs to be done and we will move forward. I am trying hard to share that outlook as well. I work with extremely bright people. Their skills definitely are better than mine and I value their input. I guess I am struggling with my ability to communicate that input and knowledge in a way that gets through to people.

I guess it's good that only 2.6 people actually read this stuff, didn't realize I would ramble this long.