Using Ontologies to Manage Resources in Grid Computing: Practical Aspects
This chapter shows how semantic technologies can enable the development of an agent-based intelligent middleware for the Grid (the Agents in Grid system). Resource providing agents are organized in teams and negotiate contracts with agents representing us
- PDF / 661,137 Bytes
- 20 Pages / 439.36 x 666.15 pts Page_size
- 46 Downloads / 200 Views
Using Ontologies to Manage Resources in Grid Computing: Practical Aspects Michał Drozdowicz, Maria Ganzha, Katarzyna Wasielewska, Marcin Paprzycki, and Paweł Szmeja
9.1 Introduction The aim of this chapter is to discuss practical aspects of the application of ontologies and semantic data processing in management of resources in the Grid. Firstly, issues involved in the development of an ontology of Grid computing are briefly considered. The discussed ontology is used not only to describe Grid resources, but also in Service Level Agreement (SLA) negotiations. Second, it is discussed how an ontology-driven user interface can be developed, to facilitate humancomputer (i.e., human-software agent) communication. Third, a solution to the problem of ontology-based agent-agent communication is presented. Finally, the role of ontologies in SLA negotiations is outlined. The chapter begins with top-level description of the system, which is used to illustrate these four main points. The Agents in Grid (AiG) project aims to develop a flexible agent-based infrastructure, which is to facilitate intelligent resource management in the Grid. Thus, the project can be considered an attempt to realize the main idea underlining the seminal paper (Foster et al. 2004), where the use of software agents as highlevel middleware for the Grid was suggested. In the AiG project, it is proposed that flexible management of resources in the Grid can be provided by teams of software agents (Kuranowski et al. 2008a,b). Furthermore, the proposed approach is based on the application of semantic data processing in all aspects of the system. Specifically, ontologies provide the metadata, to be used to describe resources, reason about them, and negotiate their usage. In addition, adaptability and flexibility of the system are to result from the application of other “agreement technologies”, being agent negotiations the most prominent one. M. Drozdowicz () • M. Ganzha • K. Wasielewska • M. Paprzycki • P. Szmeja Systems Research Institute Polish Academy of Sciences, Warsaw, Poland e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected] S. Ossowski (ed.), Agreement Technologies, Law, Governance and Technology Series 8, DOI 10.1007/978-94-007-5583-3__9, © Springer Science+Business Media Dordrecht 2013
149
150
M. Drozdowicz et al.
9.2 System Overview In the work of Wasielewska et al. (2011), the Grid is considered as an open environment (see also Chap. 7), in which Agents representing Users interact to, either (a) join a team, or (b) find team(s) to execute job(s) (Dominiak et al. 2008; Kuranowski et al. 2008a). The main assumptions behind the proposed approach were1 : • • • • •
Agents work in teams (groups of agents), Each team has a single leader—the LMaster agent, Each LMaster has a mirror, the LMirror agent that can take over its job, Incoming workers (Worker agents) join teams based on User-defined criteria, Teams (represented by their LMasters) accept Worker
Data Loading...
