This paper discusses the implementation of a distributed execution environment, DJO, which supports the use of shared Java objects for parallel and distributed applications and provides the Java programmer with the illusion of a network-wide shared object space on loosely-coupled distributed systems. DJO supports shared objects through an implementation of multiple reader/single writer write-invalidate DSM protocol in software, providing the shared memory abstraction at object granularity. Object distribution and sharing are implemented through the replication mechanism, transparently to application. The system enforces mutual consistency among replicas of an object. The main benefits of DJO are enhanced availability and performance due to the replicated object model and easier application design, as the underlying software takes care of distribution and memory consistency issues.

ABSTRACT:By combining research results from different communities, better systems can be built. This is exemplified in our designs of two mobile agent systems on top of distributed shared memory (DSM). Conventional use of DSM causes inadvertent movement of large data, which leads to the belief that DSM is not scalable. In our two designs, we use DSM in two different ways. The first uses explicit agent migration and data distribution where DSM contains only the small agent variables, while in the second all data is placed in DSM, and the agents migrate implicitly to large-sized data. The first design gives the programmers more control whereas the second design provides more transparency. The benefits of both designs include improved efficiency, ease of construction, and ease of use. I.