In part 2 of our interview with Annrai O'Toole, CEO of Cape Clear Software Inc., he answers questions about common mistakes and misconceptions about SOA. He also talks about selecting tools for SOA and what constitutes true interoperability. Finally, he looks in his crystal ball and makes some interesting, bold and also tongue-in-cheek predictions about the future. Read part one.
Now that much of the civilized world is pursuing SOA, what are the chief mistakes you see users making in 2007?
There are two mistakes. One is that people pretend to do SOA, in other words, they tell their management that what they are doing is SOA, but in reality, they keep on doing things the old fashioned way by writing lots of Java and generally viewing all IT problems as just another opportunity to write more code. Real SOA is about a radical change and thinking about IT problems as a business user rather than a techie.
The second mistake, once you get on a SOA path, is to design things that won't scale. It is critical to design a SOA application that is as uncomplicated as it possibly can be. Complex, baroque SOA is just as bad, although in a different way, than loads of Java code. What are vendors trying to sell in regards to SOA that users probably don't need?
I think the answer to that question is more one of priorities. SOA covers a huge landscape and vendors are rushing to provide a diverse set of tools to cover the entire spectrum, but there is a pretty natural progression and customers should order their product choices along that line.
In the first case, people need the basic capability to design and create services, get them to talk to one another and orchestrate them together, which is the core functions of an ESB. Then next comes the challenge around managing a set of deployed services -- both from a network and system point of view and from a coherence or governance point of view.
Then you have test and load management tools and so on. So, start small, get the basics working, and then move on to bigger things. We saw a lot of large companies gobble up smaller vendors in 2006, is it getting harder for companies like Cape Clear, Logic Library, AmberPoint and Reactivity to compete with the leviathans in the marketplace?
Leviathans are slow and not able to react well, either to market changes or the demanding needs of the majority of customers. It is a tough competitive world, but the smaller companies are using their size as an offensive weapon. We are really close to the cutting edge of the SOA world. It is our exclusive focus. We have seen the mistakes customers make and are better able to help them reap the real business benefits of SOA. What, for you, makes for true interoperability?
The real definition of true interoperability is that you just don't care what is on the far side of the wire. A browser doesn't care what Web server it uses, equally, the users of SOA services don't care what technology is used to build and host those services. This is a reality today and it is a major step forward for application interoperability.
There is one thought that's been rattling around and I think you'll see it come up in 2007. The thing we've been discussing with a lot of people is the notion of integration as a service. Obviously, software as a service has come of age and you're going to see a lot more of that. For the longest time integration has been this big problem that people have had to deal with internally, inside the four walls. If they want to integrate lots of different applications together they've got to go and buy integration technology. Maybe ESBs can do it, but in most cases they're using EAI. I believe the technology and the standards are in place to a point where you're going to see integration between different applications offered in a SOA manner as a hosted service sitting out on the Internet. This is one that's got us energized and we're looking at ways to make it a reality. Got any crazy, bold predictions for 2007?
Ireland will win the Grand Slam in the 6 Nations Rugby and also the Rugby World Cup! More seriously, BEA will finally be bought by HP.