A framework for persistence-enabled optimization of java object stores

Abstract

Aggressive optimization of programs often relies on analysis and transformation that cuts across the natural abstraction boundaries of the source programming language, such as procedures in procedural languages, or classes in class-based object-oriented languages like Java. Unfortunately, execution environments for languages such as Java dynamically link code into the application as it executes, precluding cross-cutting analyses and optimizations that are too expensive to apply on-line. Fortunately, persistent object systems usually treat the code base as an integral part of the persistent store. This code base approximates the notion of “wholeprogram” that has been exploited in other optimization frameworks. This paper describes an analysis and optimization framework for Java that operates against the persistent code base, and couples the results of analysis and optimization with the run-time system to ensure continued correctness of the resulting code. The framework performs extensive analysis over the code in the store, supporting optimizations that cut across class boundaries in ways that are not safe to perform off-line on stand-alone Java classes.