R. Qu, S. Hirano, and T. Ohkawa (Japan)
Memory analysis, Java Middleware, Distributed System
Distributed object-oriented middleware technologies have been adopted for ubiquitous communicating real-time and embedded systems. Although Java plays an important role in building distributed object-oriented middleware because of its portability and productivity, it is not popular to real-time system developers because it is slow and hard to keep real-timeness due to garbage collection. This paper presents a novel memory classification and measurement approach for modern programming language such as Java to analyze the memory utilization of middleware technologies in order to improve them for embedded systems using limited computing resource. We classified Java memory into four categories such as static, quasi-static, quasi-dynamic and dynamic, and show how to measure the size of each memory. Intensive case studies using popular Java middleware technologies such as Web Services, CORBA, RMI and HORB revealed that reducing the use of dynamic memory has contributed far more to the efficiency and performance than reducing the use of quasi-static and quasi-dynamic memory.
Important Links:
Go Back
