Providing computing service to users in a heterogeneous distributed computing environment

• US 20060190605A1
• Filed: 02/15/2006
• Published: 08/24/2006
• Est. Priority Date: 02/18/2005
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Gerber, Michael, Franz, Joachim, Spruth, Wilhelm, Gruetzner, Matthias
• Granted Patent: US 8,140,371 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: G06Q 10/0631 Resource planning, allocation, distributing or scheduling for enterprises or organisations
• CPC Class Codes: G06Q 10/06313 Resource planning in a project environment
• CPC Class Codes: G06Q 10/06315 Needs-based resource requirements planning or analysis
• CPC Class Codes: H04L 67/565 Conversion or adaptation of application format or content adding application control or application functional data H04L67/561

Abstract

For providing a service-centric approach to allocating computing service power to users in a heterogeneous distributed computing environment such as a grid, a grid service broker comprises a central service broker and distributed service brokers. Service power is measured in resource-independent service units defined and used by the service broker. The central service broker uses meta-scheduling for distributing service requests to the distributed service brokers, based on projected service units needed to process the service requests.

Claims

1. A service broker suitable for use in a heterogeneous distributed computing environment, said service broker comprising a central service broker and at least one distributed service broker, both belonging to a service provider domain and being connected to each other via a communication infrastructure;

wherein said central service broker and said distributed service broker are service centric and measure service power in service units that are resource independent and transformable in a distributed heterogeneous environment;

wherein said central service broker uses a meta-scheduling service for distributing service requests to a dedicated distributed service broker managing a single processing domain, which requests are based on projected service units needed to process incoming service requests, wherein said meta scheduling service receives collected information from the distributed service broker, said collected information comprising information about projected service units needed to process the incoming requests and remaining service power capacity in service provider processing domains;

wherein distribution of said service requests to the distributed service broker is performed by a band mapping component of said central service broker that maps incoming service requests to specific bands, each band representing specific business products with an amount of service power measured in service units with an agreed quality of service, wherein each band can be dynamically configured by a band transformer, that is part of the distributed service broker, to set individual band boundaries; and

wherein availability of service power measured in service units and the band transformer together allow resource independent and transformable economic distribution of service requests within a heterogeneous distributed computing environment.

2. The service broker according to claim 1, wherein said meta scheduling service uses an EOML3 economic optimization model scheduling policy.

3. The service broker according to claim 1, wherein said meta scheduling service is supported by an auction to establish market prices for the provided service power of individual bands.

4. The service broker according to claim 1, wherein said auction is a capacity sealed bid auction.

5. The service broker according to claim 4, wherein said distributed service broker uses an on- demand value calculation to represent processing domain specific market value for processing service requests and uses an on-demand value calculation in said capacity sealed bid auction.

6. The service broker according to claim 1, wherein said distributed service broker reserves service power according to service power requests and service power reservation requests.

7. The service broker infrastructure according to claim 6, wherein said distributed service broker reserves service power using reservation slots.

8. The service broker according to claim 1, wherein said heterogeneous environment is a grid environment.

9. The service broker according to claim 1, wherein bands are characterized by service power provided, quality of service, price and cost, dedicated resource manager, and business processes and business products supported.

10. The service broker according to claim 1, wherein said central service broker and said distributed service broker run within a service provider domain.

11. The service broker according to claim 1, wherein communication between said central service broker and said distributed service broker is provided by a grid infrastructure with TCP/IP using hyper sockets.

12. A method for providing computing service power in a distributed heterogeneous environment having a service requestor and service provider connected via a service broker, wherein the service broker distributes requests received from the service requester to the service provider, said method comprising:

receiving service requests from the service requestor;

mapping the service requests to multiple bands by the service broker;

projecting service units to be processed for the service requests on the mapped bands;

building a demand set of processing domains which have bands available sufficient for the service power needed;

requesting offers from a processing domain that is part of the demand set;

building a proposal set using received processing domain offers according to a best-fit scheduling strategy; and

accepting the proposal set to provide a reservation of service power for the accepted service request.

13. The method according claim 12, wherein mapping is based on quality of service power. needed, and on business products and business process used.

14. The method according to claim 12, wherein projecting service units is based on historical time series analysis.

15. The method according to claim 12, wherein the demand set is based on filter criteria of a service provider strategy for scheduling service requests.

16. The method according to claim 15, wherein said filter criteria are selected from:

using the cheapest processing domain, maximizing throughput, maximizing utilization of a processing domain, meeting quality of service and business service level agreements, and processing service requests using on-demand values.

17. The method according to claim 12, wherein offers are requested using a capacity sealed bid auction.

18. The method according to claim 12, wherein the proposal set is based on a best-price strategy, lowest performance index, and highest on-demand value, which are part of a service provider scheduling strategy.

19. The method according to claim 12, wherein accepting is performed by a reservation scheme that optimizes processing according to quality of service, utilization, and profit optimization targets, by smoothing peak load demands over time.

20. A computer program product for providing computing service power in a distributed heterogeneous environment having a service requestor and service provider connected via a service broker, wherein the service broker distributes requests received from the service requester to the service provider, the computer program product comprising a computer readable medium having computer readable program code tangibly embedded therein, the computer readable program code comprising:

computer readable program code configured to receive service requests from the service requester;

computer readable program code configured to map the service requests to multiple bands by the service broker;

computer readable program code configured to project service units to be processed for the service requests on the mapped bands;

computer readable program code configured to build a demand set of processing domains which have bands available sufficient for the service power needed;

computer readable program code configured to request offers from a processing domain that is part of the demand set;

computer readable program code configured to build a proposal set using received processing domain offers according to a best-fit scheduling strategy; and

computer readable program code configured to accept the proposal set to provide a reservation of service power for the accepted service request.