Blesso Consulting

Francis Blesso has helped companies solve complex techincal problems. Below is a sample of his successes.

Selected Projects

Web App Performance Re-architecture
Airline Messaging Protocols
Web Services and Message Driven Beans
Performance Tuning
Source Code Management and Deployment
Enterprise Application Integration

Web App Performance Re-architecture

Problem: The client's Spring Web Flow application was suffering performance problems from a large user increase. This content management application was consuming too much memory and cpu time. The system had become unstable and often unusable, requiring frequent restarts.
Solution: Francis identified immediate memory usage reductions to improve stability and assisted in deploying a quick-fix release. He designed a new architecture and introduced architectural best-practices to reduce memory usage, improve response time, and further increase stability. Francis tuned the Hibernate mappings to improve data-access performance. Instrumentation showed a decrease in response times from tens of seconds to under one second in the final release. Bug submissions from that release decreased by 90 percent.

Airline Messaging Protocols

Problem: For testing an airline reservation system, the client needed to convert between ASCII-formatted industry messages and "Plain Old Java Objects." These teletype message formats are decades old with tricky documentation. The conversion process was overwhelming the team and Francis helped temporarily.
Solution: Francis rapidly converted several message formats, creating tests to validate the round-trip conversion from thousands of actual, archived messages. Francis completed the conversion so quickly and precisely, that the team manager asked him to join the team and kept him for nearly four years where he became the team's technical leader.

Web Services and Message Driven Beans

Problem: To price policies, a large property insurer needed to determine the FEMA flood risk. A third party provides a FEMA flood risk web service over the internet. However, the client did not want to wait for the lookup.
Solution: Francis created a Message Driven Bean to handle the flood zone lookup. When the user saves address information, the system writes a JMS message containing the latitude and longitude. The Message Driven Bean handles the message by invoking a Web Services request using Apache Axis to look up the flood zone. Upon determining the flood zone, the MDB updates the address record with the flood zone value. Built-in retry logic ensures no request is lost due to service outages or other exceptions.

Performance Tuning

Problem: A client's premium management batch program was taking more than 24 hours to process a large policy. Several client attempts failed to improve performance.
Solution: Francis analyzed the code, identified several poorly performing algorithms, and coded faster versions. The re-written program processed in one-half hour, well within the batch window.

Source Code Management and Deployment

Problem: A Fortune 500 company needed to manage custom code and vendor code for an e-procurement package. None of the team members had any prior source code management experience.
Solution: Francis designed an approach to link the custom package code with CVS while respecting the vendor's file structure. A CruiseControl implementation provided developers the capability to deploy the code and track builds. He provided training for CVS, the Eclipse IDE, CruiseControl, and related Unix scripts.

Enterprise Application Integration

Problem: To evaluate underwriting performance, a leading workers compensation provider needed to measure a policy's premium versus a standard benchmark premium. The underwriting system would compute premiums using the policy rates and the benchmark rates, and send both premium sets to a data warehouse.
Solution: Francis worked with business owners to document, analyze, and refine the requirements. From the requirements, he defined an XML format to exchange premium data between the underwriting system and the data warehouse. Francis built the component to create the XML and send to the data warehouse system via MQSeries.