Abstract:
The growing number of middleware for building distributed systems has made it increasingly difficult to monitor and manage them. This is especially the case when different and often interacting systems based on different middleware technologies are deployed on a single site. Proper administration of such systems requires proper monitoring, which in turn requires the extraction of quality information from the underlying systems at runtime. The proxy pattern, although seemingly simple, can be put to good effect to address problems of this nature. By dynamically generating proxies capable of intercepting and logging invocations the characteristics, behavior and relationships of processes can be ascertained at runtime, providing a global view on the behavior of entire systems in operation. This paper addresses these two concerns, the need to have a comprehensive global view on distributed systems and the viability of the proxy pattern as a technique for achieving it.
