Horizontal Services Provided by EJB

Traditional Buy Versus Build Analysis

One of the major decisions that must be made when building any nontrivial application is when to buy a set of components or develop them in-house. With most components , this might be an easy decision. When you have requirements for something that is unique to your software solution and you think your organization knows the domain better than anyone else, this is normally a build situation. Or when the cost is so high to buy a solution, sometimes it makes better economical sense to have developers build what you need.

The decision might be totally different if we were to consider something really complex. Some components are better off left to the experts to do because you would end up spending more time and money trying to build one. A perfect example is an EJB server. You typically don't want to build one from scratch. The amount of time and expertise needed exceeds most development shops .

Horizontal services, on the other hand, are components that you normally want to build unless you can find ones that closely match your requirements. The expertise curve is not as high as, say, building an EJB server. In many cases, it makes economical sense to build what you need rather than try to integrate and trim down from too much functionality that might be provided by a third-party solution.

Even when a buy decision is made, this doesn't mean that design or implementation is not necessary. You still have to integrate the product into your application. In some cases, this integration can be worse than building something from scratch. However, by having specifications and contracts between clients and the service APIs, the integration gets much easier.

Lately, there has been a rash of open -source projects putting together everything from EJB servers to logging APIs. It might also be worth your while to invest some time in evaluating these, in addition to the traditional buy versus build analysis process.