Displaying cache information in the JADE object oriented distributed processing system

Content Partner: Lincoln University. One of the major factors affecting the performance of any distributed processing system is the management of the cache in the local workstation. JADE is an object oriented distributed processing system developed by Aoraki Corporation of Christchurch, New Zealand. JADE uses a fully object model both for its underlying database and its development environment. Much of JADE is itself written in the JADE language. A distributed transaction processing system such as JADE must on the one hand provide for full data integrity using appropriate locking and database management techniques and on the other hand ensure that local cache is managed in such a way that users experience the best performance possible and the network is not unduly loaded. This is made more complex because the flexibility of JADE means that application code (methods) can run on either the client or the server depending on where the developer perceives it will run more efficiently. The basic unit of caching in JADE is an individual object. Each object occupies a buffer in the cache memory. At present JADE uses a “least recently used” (LRU) cache-ordering algorithm (Stallings, 1993, p. 163) in which the last cache buffer referenced is placed at the top of the cache. When space is required for a new buffer those at the bottom are displaced first. This paper is an initial attempt at providing a visual description of the arrangement of objects in the cache, based on snapshots of cache information taken while a JADE application is run.