Concurrency: A Case Study in Remote Tasking and D istributed I

Remote tusking encompasses different functionality, such as remote forking, multiple remote spawning, and task migration. In order to overcome the relatively high costs of these mechunisms, optimizations can be upplied at various levels of the underlying operating system or application. Optimizations include concurrent message transmission, increased throughput and reduced latency at the distributed IPC level; batching, overlapping, andpipelining at the remote tasking level; and multithreading at the application level. Of purticular interest is the resulting concurrency, since in a complex program, it may be a dominant performance factor. Distributed IPC is typicully characterized by throughput and latency. However, many design and implementation details importunt for reul application performance remain unobserved by this simple characterization. This puper describes distributed IPC from u remote tasking point of view. Remote tasking exercises all aspects of distributed IPC extensively. We analyze two versions of distributed IPC supported in Mach (NORMA IPC and DIPC).