MINIMIZING COMPLEX DECISIONS TO ALLOCATE ADDITIONAL RESOURCES TO A JOB SUBMITTED TO A GRID ENVIRONMENT

• US 20080256228A1
• Filed: 06/21/2008
• Published: 10/16/2008
• Est. Priority Date: 01/13/2004
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: HAMILTON, RICK ALLEN, JOSEPH, JOSHY, FELLENSTEIN, CRAIG, SEAMAN, JAMES
• Granted Patent: US 8,387,058 B2
• Time in Patent Office: Days
• US Class Current: 709/223
• CPC Class Codes: G06F 9/50 Allocation of resources, e.g. of the central processing unit [CPU]

Abstract

A method, system, and program for minimizing complex decisions to allocate additional resources to a job submitted to a grid environment are provided. First, at least one decision to allocate additional resources to at least one previously submitted job is stored in a decision cache, wherein said the least one decision is stored according to at least one characteristic of the at least one previously submitted job. When another job is submitted to the grid environment needs additional resources, the characteristic of the currently submitted job is compared with the characteristics of previous submitted jobs. If there is a match, then the previously made decision associated with the matching characteristic controls allocation of additional resources for the currently submitted job, such that complex decision making for allocation of additional resources is minimized by reusing previously stored decisions to allocate additional resources.

Claims

1. :

A system for minimizing complex decision making when allocating additional resources to a job submitted in a grid environment, comprising;

a grid environment comprising a plurality of computing systems each comprising at least one resource and communicatively connected over a network through a grid management system to share each said at least one resource through at least one web service layer atop at least one grid service layer implemented within an open grid services architecture;

said grid management system for receiving a plurality of separate jobs from a plurality of client systems over said network;

said grid management system for managing distribution of each of said plurality of separate jobs to a separate selection of said at least one resource;

said grid service layer, responsive to said grid management system distributing each of said plurality of separate jobs to said separate selection of said at least one resource, for triggering at least one resource monitor to monitor a performance of said separate selection of said at least one resource in handling each of said plurality of separate jobs to meet a separate performance agreement specified for each of said plurality of separate jobs and triggering at least one decision controller from said at least one grid service layer to determine whether to allocate additional resources to each of said plurality of separate jobs to meet at least one said separate performance agreement specified for at least one of said plurality of separate jobs;

a memory medium within said grid management system, responsive to said at least one decision controller deciding to allocate additional resources to at least one particular job from among said plurality of separate jobs, for storing a record of at least one decision to allocate additional resources from among said at least one resource to said at least one particular job, wherein said record of said at least one decision is specified by an amount of said at least one resource available for allocation and at least one characteristic of said at least one particular job, wherein said at least one characteristic of said at least one particular job comprises a system state detected by said at least one resource monitor of a particular separate selection of said at least one resource to which said at least one particular job was submitted and a particular priority of said at least one particular job to receive additional resources from among said at least one resource;

said grid management system, responsive to said at least one decision controller determining whether to allocate additional resources to a currently submitted job from among said plurality of separate jobs, for comparing at least one characteristic of a currently submitted job from among said plurality of separate jobs qualifying for additional resources from said at least one resource of said grid environment with said at least one characteristic of said at least one particular job in said record in said memory medium, wherein said at least one characteristic of said currently submitted job comprises a system state of a current separate selection of said at leas tone resource to which said currently submitted job was submitted and a current priority of said currently submitted job to receive additional resources from among said at least one resource;

said grid management system, responsive to said at least one characteristic of said currently submitted job matching said at least one characteristic of said at least one particular job in said record, for applying said record of said at least one decision associated with said matching job characteristics to control allocation of additional resources for said currently submitted job, such that complex decision making for allocation of additional resources is minimized by reusing previously stored decisions to allocate additional resources.

2. :

The system according to claim 1 for minimizing complex decision making wherein said memory medium comprises a table for storing said record of said at least one decision to allocate additional resources to each of said at least one previously submitted jobs, each stored according to said at least one characteristic.

3. :

The system according to claim 1 for minimizing complex decision making wherein said grid management system further comprises;

means, responsive to said at least one characteristic of said currently submitted job not matching said at least one characteristic of said at least one particular job, for deciding whether to allocate additional resources to said currently submitted job and storing an additional record for said decision according to said characteristic of said currently submitted job; and

means, responsive to a decision to allocate additional resources to said currently submitted job, for deciding which resources to allocate and storing said additional record for said decision according to said characteristic of said currently submitted job.

4. :

The system according to claim 1 for minimizing complex decision making wherein said grid management system further comprises;

means for organizing a plurality of discrete sets of physical resources into a hierarchy of a plurality of logical resources within said grid environment, wherein said additional resources are allocated from said hierarchy of said plurality of logical resources.

5. :

A computer program product, comprising a volatile or non-volatile machine-readable medium including instructions for minimizing complex decision making when allocating additional resources to a job submitted in a grid environment, wherein said computer-readable program when executed on a computer causes the computer to;

enable a grid environment comprising a plurality of computing systems each comprising at least one resource and communicatively connected over a network through a grid management system to share each said at least one resource through at least one web service layer atop at least one grid service layer implemented within an open grid services architecture;

receive a plurality of separate jobs from a plurality of client systems over said network at said grid management system;

manage distribution from said grid management system of each of said plurality of separate jobs to a separate selection of said at least one resource;

responsive to said grid management system distributing each of said plurality of separate jobs to said separate selection of said at least one resource, trigger at least one resource monitor from said at least one grid service layer to monitor a performance of said separate selection of said at least one resource in handling each of said plurality of separate jobs to meet a separate performance agreement specified for each of said plurality of separate jobs and triggering at least one decision controller from said at least one grid service layer to determine whether to allocate additional resources to each of said plurality of separate jobs to meet at least one said separate performance agreement specified for at least one of said plurality of separate jobs;

responsive to said at least one decision controller deciding to allocate additional resources to at least one particular job from among said plurality of separate jobs, store, in at least one memory medium of said grid management system, a record of at least one decision to allocate additional resources from among said at least one resource to said at least one particular job, wherein said record of said at least one decision is specified by an amount of said at least one resource available for allocation and at least one characteristic of said at least one particular job, wherein said at least one characteristic of said at least one particular job comprises a system state detected by said at least one resource monitor of a particular separate selection of said at least one resource to which said at least one particular job was submitted and a particular priority of said at least one particular job to receive additional resources from among said at least one resource;

responsive to said at least one decision controller determining whether to allocate additional resources to a currently submitted job from among said plurality of separate jobs, compare, by said grid management system, at least one characteristic of a currently submitted job from among said plurality of separate jobs qualifying for additional resources from said at least one resource of said grid environment with said at least one characteristic of said at least one particular job in said record in said at least one memory medium, wherein said at least one characteristic of said currently submitted job comprises a system state of a current separate selection of said at leas tone resource to which said currently submitted job was submitted and a current priority of said currently submitted job to receive additional resources from among said at least one resource;

responsive to said at least one characteristic of said currently submitted job matching said at least one characteristic of said at least one particular job in said record, apply by said grid management system said record of said at least one decision associated with said matching job characteristics to control allocation of additional resources for said currently submitted job, such that complex decision making for allocation of additional resources is minimized by reusing previously stored decisions to allocate additional resources.

6. :

The computer program product of claim 5, wherein said computer-readable program when executed on a computer further cause the computer to store said record of said at least one decision to allocate additional resources to each of said at least one previously submitted jobs, each stored according to said at least one characteristic in a table of said memory medium.

7. :

The computer program product of claim 5, wherein said computer-readable program when executed on a computer further cause the computer to;

responsive to said at least one characteristic of said currently submitted job not matching said at least one characteristic of said at least one particular job, decide whether to allocate additional resources to said currently submitted job and storing an additional record for said decision according to said characteristic of said currently submitted job; and

responsive to a decision to allocate additional resources to said currently submitted job, decide which resources to allocate and storing said additional record for said decision according to said characteristic of said currently submitted job.

8. :

The computer program product of claim 5, wherein said computer-readable program when executed on a computer further cause the computer to organize a plurality of discrete sets of physical resources into a hierarchy of a plurality of logical resources within said grid environment, wherein said additional resources are allocated from said hierarchy of said plurality of logical resources.