The next old thing - SOA

This entry was inspired by the article The spiral staircase of SOA.

I find the latest SOA zeal both comforting and amusing. Several of us that were successful in building and deploying CORBA based distributed applications realized early on that using an OO methodology was generally doomed to failure. Many "failed" CORBA projects were because people tried to apply OOD to CORBA interfaces and the completely forgot about the fact that there was -or might be in the future - a WAN between their clients and servers or providers and consumers. Fine grained interfaces gave CORBA vendors a bad name and unfortunately there weren't enough "wizards" (smart developers) to go around. Service based interfaces forced designers to think more about essential contracts between their applications and those applications users/consumers.

I predict however that there will be similar failures in future SOA based projects - though perhaps less due to the emphasis on "service". The problem has not been the technology - for the last decade anyway (with CORBA and Java technologies like J2EE). The problem is that people continue to be poorly educated on best practices - on what a service is and why it's important to remember that though your application may "close" to it's consumer today it might not be tomorrow. Therefore consideration of a WAN is important. E.g. a good retail business understands that though their service might be exceptional they must also consider location. Location, location, location. Modern technologies do such a great job of location transparency that developers often program as if the service (in their view the "object") is right there in memory, close by.

That's why the word "Service" in SOA is so important and the word "Object" in CORBA was so dangerous. Hopefully people will take note.

Another thing to note based the article is the middleware fashion trend. I'll give you the short version - as a technology trend matures and gains wider usage, developers and architects usually encounter the complexities of enterprise features. As developers and architects struggle with these complexities someone else comes along with something new and innovative. This is not a bad thing. They also demonstrate how "easy" it is with a Hello World demo or something similar. Everyone applauds and jumps into the technology (for various reasons depending on who they are and their skill levels ;-). Then they hit the complexities of enterprise features, they get frustrated, and they start to look bad. And thank heavens for the new and improved technology that appears just when they need it (BTW have you seen its "Hello World" demo?).... and off the cycle goes again.

I understand that it is more complicated than that and that there have been great innovations. But there is merit in the argument that we are reinventing lots of stuff over and over again because we fail to master existing technology and improve it.

BTW thanks to Jon Udell for encouraging me to get this up on a blog. IP Babble was born.