Project Summary: A Benefits & Insurance Management Corporation
Background
This company is a leading provider of life insurance and benefit plan management solutions. For larger customers who are interested in providing their own user interface, they provide backend access and support for direct access to the underlying management system. For their database management system, they have used Progress to maintain large customer data repositories, and provided COM+ and HTML gateways to their backend data. However, upon acquiring one of the world's leading financial management and advisory companies as a client, this corporation needed to provide a solution based on MQ Series, which their client uses as a standard messaging platform.
Solution
One of SDI’s partners provided key assistance with the architecture of a new data access solution that tied together MQ Series running on Linux to the Progress Application Server running on Windows 2000 by means of a Java messaging daemon also running on Linux.
- Requirements
- Message Design
- Writing and Developing the Application
- Training and Basic Application Design
- Failover
The financial management and advisory client intended to provide the data retrieved from the corporation's system to the users on their benefit plan website thus requiring an end-to-end time of less than 5 seconds. Also, the system had to be redundant and provide failover capabilities.
XML messages were designed for relaying complex user data for both plan enrollment and plan review. The message design was codified using XML Schema.
The application itself is a middleware daemon that pulls messages from MQ Series, parses the messages, and dispatches them to the appropriate Progress function handlers available at the Progress application server. The daemon is multi-threaded allowing for high volume processing. The daemon also uses two XSL Transform caches to reformat messages from hierarchical XML to flat for inbound processing and from flat to hierarchical for outbound processing. Between judicious caching and multi-threading, even at high load the average response time was less that .5 seconds allowing for the website response times to be met.
Following the design and implementation of the system, documentation was written and 2 days of training were provided to enable the internal support staff to manage and monitor the daemon performance.
In order to provide the highest degree of uptime, a Linux cluster was built using the Linux High-Availability (HA) package and heartbeat.
About the Application
The application is currently in use providing enrollment and benefit plan management services on a 24x7x365 basis to one of the world’s financial giants.