4. Migration / Implementation

The migration of the application to OccamJ took approximately 6 weeks duration. The migration task involved removing EJB framework dependencies and re-mapping of the entities and relationships within the Open Access Workbench.

The migration task was made significantly easier because EJB specific code such as context lookups, finders etc were factored out into helper classes and standard code patterns were used throughout the application. Many of the changes were made through text search-andreplace techniques.

5. Results

The results exceeded expectation.

5.1. Number of implemented classes

Removal of EJB generated implementation classes reduced the number of server classes from over 3000 down to approx. 800.

5.2. Deployment descriptors

The size of the deployment descriptors for the system was reduced dramatically from approx. 50,000 lines to less than 1000.

5.3. Compile and code deployment times

Compile times for the server code were reduced from 7mins 40 sec. down to 17 sec. Deploy time was reduced from 1 min 40 sec down to 2 sec.

5.4 Production performance

After migration the application ran approx. 15% faster on OccamJ than the EJB server. In addition, tuning of selected functionality using Open Access JDO allowed much greater performance gains in these critical areas.

6. Conclusion

Adoption of a POJO-POJO architecture for enterprise-scale server applications can greatly reduce complexity, improve developer productivity and enhance runtime performance. OccamJ with Open Access JDO deliver these benefits in real-world applications.

Your Name / Email Address
Comment
Spam Protection - Please enter the code in the image -

Listen to code