Distributed Shared Memory: A Review

  • Author(s) / Creator(s)
  • Technical report TR96-22. There are two common programming models for intercommunication in parallel and distributed applications: shared memory and message passing. Shared memory has been the standard model for tightly-coupled systems (multiprocessors), where the processors have uniform access to a single global memory. Although it is easy to use, memory contention limits the scalability of systems using this model. On the other hand, message passing has been the major model for loosely-coupled systems (multicomputers), where each processor has a physically separate private memory. Applications that use message passing have to move the data back and forth explicitly within programs making the model burdensome for programmers. In recent years, researchers exploited the shared memory paradigm and studied its applicability to loosely-coupled systems. These efforts resulted in the introduction of a new concept, the abstraction of shared memory on a distributed system, that combines the best of the two original models. This concept, commonly known as \"Distributed Shared Memory (DSM)\", provides the illusion of a large ``shared'' memory that extends across machine boundaries. This paper reviews current research in distributed shared memory and related topics and summarizes the research results in the field. | TRID-ID TR96-22

  • Date created
  • Subjects / Keywords
  • Type of Item
  • DOI
  • License
    Attribution 3.0 International