Summary: | In this paper we describe a new model for programming parallel and distributed systems, called the nested object model. It extends the shared data object model and solves an important problem with this model. 1 The Shared Data Object Model Orca is a language for parallel programming on distributed systems, based on the shared data-object model [Bal91, Bal92]. Processes in Orca can share variables (data-objects) of abstract data types (ADTs). Each ADT operation is applied to a single object and is executed atomically. This model can be implemented efficiently on a distributed system by replicating objects with a high read/write ratio. Read operations are executed on the copy in local memory, while write operations are broadcast, using totally-ordered group communication [Kaashoek92] to ensure sequential consistency. The model is a simple and portable form of object-based distributed shared memory. 2 Problems with the Shared Data Object Model Experience with Orca taught us that the maj.
|