Title :
A resource allocation architecture with support for interactive sessions in utility Grids
Author :
Talwar, Vanish ; Agarwalla, Bikash ; Basu, Sujoy ; Kumar, Raj ; Nahrsted, Klara
Author_Institution :
Hewlett Packard Labs., Palo Alto, CA, USA
Abstract :
Utility Grids implement a virtualization architecture and allow for sharing of infrastructure for improved return on investment (ROI). We consider extending the existing Grid infrastructure to support interactive sessions in an enterprise setting. This would allow users to remotely execute interactive applications in the Grid and view their output using remote display technologies. We propose a resource allocation architecture to support such interactive sessions in utility Grids. End-users submit requests for a remote desktop session. A utility Grid site resource allocation system assigns a compute node for this request considering performance needs of such remote desktop sessions. We also propose a hierarchical admission control system. The system consists of a site admission control system for admission control of remote desktop sessions, and a session admission control system at the compute node for admission check of per-application interactive sessions. We also present discussion on mixed workloads consisting of requests for batch jobs and interactive remote desktop sessions.
Keywords :
business data processing; grid computing; interactive systems; performance evaluation; resource allocation; scheduling; telecommunication congestion control; workstation clusters; ROI; batch job requests; enterprise setting; hierarchical admission control system; interactive remote desktop sessions; interactive sessions; mixed workloads; performance; remote desktop session; return on investment; site admission control system; site resource allocation system; utility Grids; virtualization architecture; Admission control; Application virtualization; Computer architecture; Computer displays; Electronic mail; Grid computing; Laboratories; Processor scheduling; Resource management; Resource virtualization;
Conference_Titel :
Cluster Computing and the Grid, 2004. CCGrid 2004. IEEE International Symposium on
Print_ISBN :
0-7803-8430-X
DOI :
10.1109/CCGrid.2004.1336705