Harnessing Replication in Object Allocation

The design of distributed systems involves the partitioning of the system into components or partitions and the allocation of these components to physical nodes. Techniques have been proposed for both the partitioning and allocation process. However these techniques suffer from a number of limitations. For instance object replication has the potential to greatly improve the performance of an object orientated distributed system but can be difficult to use effectively and there are few techniques that support the developer in harnessing object replication. This paper presents a methodological technique that helps developers decide how objects should be allocated in order to improve performance in a distributed system that supports replication. The performance of the proposed technique is demonstrated and tested on an example system.

Object Allocation with Replication in Distributed Systems

The design of distributed systems involves dividing the system into partitions (or components) and then allocating these partitions to physical nodes. There have been several techniques proposed for both the partitioning and allocation processes. These existing techniques suffer from a number of limitations including lack of support for replication. Replication is difficult to use effectively but has the potential to greatly improve the performance of a distributed system. This paper presents a new technique technique for allocating objects in order to improve performance in a distributed system that supports replication. The performance of the proposed technique is demonstrated and tested on an example system. The performance of the new technique is compared with the performance of an existing technique in order to demonstrate both the validity and superiority of the new technique when developing a distributed system that can utilise object replication.