Improving the Currency of Information in Large-Scale Grid Information Systems

  • Laurence Field

Student thesis: Phd


Grid computing aims to seamlessly deliver computing power as a resource similar to how electrical power is delivered over the electrical power grid.This thesis studies the scalability of Grid information systems, which enable the discovery of resources in a Grid computing infrastructure and provide further information about their structure and state.It is motivated by the observation that existing approaches for query processing in a Grid environment do not take into consideration the dynamic nature of the information.The concept of static and dynamic information has been widely used in Grid computing to describe the information about resources even though a detailed definition does not exist and its use may be misleading. This work claims that all information is dynamic and that it is the expected frequency of change which defines the relative dynamic nature of information types.As Grid infrastructures increase in size, the volume of information describing all the resources in the infrastructure increases and hence the number of changes due to its dynamic nature will also increase. Since not all information changes at the same frequency, one can use a measurement that defines the freshness of information to optimise a Grid information system.Ultimately, the freshness is maximized if each change is propagated through the system as soon as it occurs at the resource, and hence the time taken to propagate this information, the latency, is minimized. This thesis proposes a novel Grid information system architecture that reduces the latency for dynamic information by sending each change as it occurs.The proposed architecture not only improves the system's ability to manage dynamic information, but in addition improves the efficiency of the system by reducing both the network throughput and operational overhead. The proposed architecture is evaluated by considering the real use cases from the Worldwide LHC Computing Grid, and is compared against the performance results of the system currently used in production today.The proposed architecture is shown to significantly improve the information freshness for dynamic information.
Date of Award1 Aug 2014
Original languageEnglish
Awarding Institution
  • The University of Manchester
SupervisorRizos Sakellariou (Supervisor)

Cite this