A model for the coordination of mobile processes

Abstract

Coordination is an important part of any computer program. As well as performing computation, applications need to interact with their environment (for example, reading a file from a disk, displaying information to a user). In distributed applications, processes that together make up the overall program also need to coordinate amongst themselves. Such coordination includes synchronisation and exchange of data.

Allowing processes to be mobile, where they are able to migrate around a network of nodes, allows novel solutions to be programmed where alternatives with static networks of processes would be inefficient or impossible to implement. For example, one method of finding "interesting" data on a network would be for a client machine to read the data off a variety of servers, and perform the filtering locally. A mobile solution could be to migrate processes to the servers, where they would query the data locally, migrating back to the client with the results. If the filter processes are small, and the databases large, this would result in reduced network usage and decreased search time.

Coordination becomes problematic in the presence of mobility. If two processes are allowed to migrate around a network, it becomes difficult to maintain connectivity between them.

This thesis explores how a set of interacting mobile processes may coordinate their activities. By abstracting away from physical networks to a model based on channels, distributed applications designers need not be aware of where processes are executing; two processes will always be able to communicate if they have access to a common channel.

More information

Identifiers

Catalogue record

Export record

ASCII CitationAtomBibTeXData Cite XMLDublin CoreDublin CoreEP3 XMLEndNoteHTML CitationHTML CitationHTML ListJSONMETSMODSMPEG-21 DIDLOpenURL ContextObjectOpenURL ContextObject in SpanRDF+N-TriplesRDF+N3RDF+XMLRIOXX2 XMLReferReference ManagerSimple Metadata