Java Tutorial - Java Scipt :
Container-Managed Persistence 1.0
CMP beans are meant to present a shared, object-oriented view into persistent data, notably relational databases. In many ways, CMP 1.0 is a naive approach that merely maps the object interface to the data source. By attempting to be minimalist, key features were omitted from the specification. Among the list of problems are:
· No ability to define relationships between CMP beans
· Limited query options
· No ability to order results
· No way to model fine-grained objects
· Huge disconnect from proven RDBMS (relational database management system) concepts
The list goes on and on. For applications that have the luxury of creating brand-new database schemas and need an EJB-centric view of the data, CMP 1.0 can be made to work. There are also a number of design patterns to work around some of its shortcomings. These days, CMP 1.0 is an outdated, very simplistic, lowest-common-denominator data mapping. In the grand scheme of things, CMP 1.0 never really took off, and because it shouldn’t be used in newer projects, will likely fade away quite soon.