Network multicomputing using recoverable distributed shared memory

A network multicomputer is a multiprocessor in which the processors are connected by general-purpose networking technology, in contrast to current distributed memory multiprocessors where a dedicated special-purpose interconnect is used. The advent of high-speed general-purpose networks provides the impetus for a new look at the network multiprocessor model, by removing the bottleneck of current slow networks. However, major software issues remain unsolved. It is pointed out that a convenient machine abstraction must be developed that hides from the application programmer low-level details such as message passing or machine failures. Use is made of distributed shared memory as a programming abstraction, and rollback recovery through consistent checkpointing to provide fault tolerance. Measurements of the authors´ implementations of distributed shared memory and consistent checkpointing show that these abstractions can be implemented efficiently.<>