Keywords: software agents, grid, resource brokering and management, ontology, semantic information processing.

Our research (the Agents in Grid; AiG project) concerns the development of an agent-based Grid middleware, in which (a) agents work in teams (each team is to be managed by the LMaster agent), (b) all meta-information is ontologically demarcated and semantically processed (with all team information stored in and managed by the Client Information Center; CIC infrastructure represented by the CIC agent), and (c) an economic model is to be based on autonomic Service Level Agreement (SLA) negotiations and Quality of Service (QoS) monitoring.

In the proposed system, two main use case scenarios provide the background against which the AiG Ontology was developed. The first scenario involves a User wishing to make her resource available for use by a team. The second scenario describes a case of a User who wishes to have a job executed using resources managed by one of the teams.

For these two scenarios, ontologies have to provide the necessary vocabulary to describe all needed concepts - both on the level of inter-agent communication, and for storing information in agents' knowledge bases. Here, notice first that both scenarios rely heavily on concepts describing resources, or resource requirements. This common ontology is named an AiG Grid Ontology and is an extension and modification of the CoreGRID ontology. Second, in both cases there is a need for concepts used in contract negotiations and related to contract conditions (the Service Level Agreement) between the participating parties (agents). This constitutes the AiG Conditions Ontology. The last remaining part of the AiG Ontology is a set of classes representing messages exchanged by the agents - the foundation of the communication and negotiation protocols in the system.

The aim of this chapter is to present the AiG Ontology, which was developed for the purpose of our system. Here, let us stress that the proposed ontology has been developed with the complete agent-based resource brokering system in mind. Therefore, it not only extends concepts that are purely Grid-based (that are only a slight modification and extension of the CoreGRID ontology), but also adds a complete set of concepts and classes that make autonomous negotiations leading to an SLA formation possible.

Finally, let us note that in our work we were primarily concerned with computational Grids (as the main application area for Grid computing of today) rather than, for instance, data Grids. However, the proposed ontology should be applicable also to the other types of Grid almost without modifications.

purchase the full-text of this chapter (price £20)

go to the previous chapter
go to the next chapter
return to the table of contents
return to the book description
purchase this book (price £95 +P&P)