Extending a persistent object framework to enhance enterprise application server performance
John Grundy, Steve Newby, Thomas M Whitmore, Peter Grundeman · Swinburne figshare (Swinburne University of Technology) · 2002
High-volume transaction processing speed is critical for adequate performance in many enterprise application servers. We describe our experiences using an object-oriented persistency framework to achieve greatly enhanced server response by the transparent use of main-memory database technology. We took an application server whose data persistency is abstracted via a persistent object framework and replaced a version of the framework using a relational database for persistency with one that uses a memory database. No changes to any of the application server components were necessary to achieve this and we achieved between 10-20 times transaction processing performance improvement. We briefly discuss some extensions to our memory database and mapping framework necessary for large-scale enterprise systems support and for data-oriented systems integration. We hope our experiences will be useful for others, both in terms of techniques for abstracting object persistency mechanisms and in approaches to application server performance enhancement.