Browse Prior Art Database

Workload Management for Service Brokers in a Grid

IP.com Disclosure Number: IPCOM000028510D
Original Publication Date: 2004-May-18
Included in the Prior Art Database: 2004-May-18
Document File: 2 page(s) / 44K

Publishing Venue

IBM

Abstract

This article describes a new kind of workload management and resource allocation system to enable a brokerage system in which resources are pre-allocated on a short-term basis to match required load.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 2

Workload Management for Service Brokers in a Grid

For certain types of application, flexible/dynamically organised distributed computing systems can offer, the opportunity for brokering of IT resources. The broker buys resources cheaply, adds value and sells the resources and/or services at a premium to clients. The model depends on standardised systems for communication, such as Web Services, and standard models for economic interaction between the various parties.

    Current workload management systems focus on optimising the use of a fixed set of resources measured against desired Service Levels. A new kind of workload management and resource allocation system is needed to enable a brokerage system.

    The current state of the art concerning Grid Economic services can be found at the following locations which are owned by the Global Grid Forum. Note the Use Cases document which contains a description of the Broker scenario.

Grid Economic Services Architecture Working Group (GESA-WG) website [http://www.doc.ic.ac.uk/~sjn5/GGF/gesa-wg.htm] Grid Economic Services Architecture (GESA) [http://www.doc.ic.ac.uk/~sjn5/GGF/CompEconArch-GGF7.pdf] Grid Economic Use Cases (GESA-UC) [http://www.doc.ic.ac.uk/~sjn5/GGF/draft-ggf-gesa-use-cases-01-7.pdf]

    In the new kind of workload management and resource allocation system resources are pre-allocated on a short-term basis to match required load. The capacity of the resources is expressed in terms of service execution rate over a fixed period of time with commitments to certain Service Levels. Such pre-allocations may come from multiple sources. They represent committed physical resources managed by the service supplier. The broker distributes work amongst the suppliers according to their relative capacities, choosing the most appropriate supplier for the service level of any particular request.

    The short-term nature of the resource allocation constrains the kind of application which can exploit the system: the effects of the workload management choices made by the broker must be predominantly contained within the period of pre-allocation. Ho...