SYSTEM, METHOD AND PROGRAM PRODUCT FOR COST-AWARE SELECTION OF TEMPLATES FOR PROVISIONING SHARED RESOURCES

• US 20130138812A1
• Filed: 11/25/2011
• Published: 05/30/2013
• Est. Priority Date: 11/25/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Assuncao, Marcos Dias De, Netto, Marco Aurelio Stelmar, Peterson, Brian L., Renganarayana, Lakshminarayanan, Rofrano, John J., Young, Christopher C.
• Granted Patent: US 8,930,541 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: G06F 9/50 Allocation of resources, e.g. of the central processing unit [CPU]
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A template generator organizing templates in a cost-aware clustering, a method of allocating resources using cost-aware clustering and computer program products therefor. A resource provisioning unit generates, selects and maintains a selected number of resource templates. Each template specifies an allocable resource capacity configuration. Each requesting client device has resources allocated determined by one of the selected resource templates. A resource provisioning unit includes a configuration store with costs of allocable resources and associated attributes, a server request store with previously received requests, and an input parameter store with template list options. A template generator determines an optimum list of templates to satisfy previously received requests. A template store stores generated template lists.

Claims

1. A shared resource system comprising:

a plurality of client devices;

one or more service provider computers selectively making resource capacity available to said plurality of client devices;

a network, resource capacity being provided to said plurality of client devices from said one or more service provider computers over said network; and

a resource provisioning unit generating, selecting and maintaining a selected number of resource templates, each template specifying an allocable resource capacity configuration, each requesting client device having resources allocated determined by one of the selected resource templates.

2. A shared resource system as in claim 1, wherein said resource provisioning unit comprises:

a configuration store storing costs of allocable resources and associated attributes;

a server request store storing previously received requests for capacity;

an input parameter store storing template list options;

a template generator determining an optimum list of templates to satisfy said previously received requests; and

a template store storing generated template lists.

3. A shared resource system as in claim 2, wherein said template generator generates a template list for each service provider computer, each generated said template list being stored in said template store.

4. A shared resource system as in claim 2, wherein said template list options comprises a complete list option and a top k list option.

5. A shared resource system as in claim 4, wherein said top k list option indicates a plurality of values for k, said template generator generating template lists for all values of k for said each service provider computer, one of the generated said template lists being selected for said each service provider computer and stored in said template store.

6. A shared resource system as in claim 5, wherein each selected template list has a minimum cost of said template lists generated for said each service provider computer.

7. A shared resource system as in claim 2, wherein each said template specifies an allocable configuration comprising a number of cores, memory space and disk space.

8. A method of allocating resources, said method comprising:

providing a collection of requests for resource capacity from a plurality of computers;

associating costs with allocable computer resources;

matching each request with at least one template;

selecting the template having a minimum cost;

sorting minimum cost templates according to the number of matched requests; and

determining the cost of matching requests for a selected number (k) of most frequently matched templates, the k most frequently matched templates being listed in a top k list.

9. A method of allocating resources as in claim 8, wherein matching said each request comprises:

generating a list of resource match templates, said list listing each template that matches at least one request;

associating a cost with each listed template; and

determining from all templates with associated costs a matching template for each request with minimum cost.

10. A method of allocating resources as in claim 9, wherein determining said matching template with minimum cost comprises:

finding a template with a minimum resource match for each request, said template being a minimum match template; and

determining the cost for resources for said minimum match template.

11. A method of allocating resources as in claim 10, wherein determining top k lists comprises:

listing sorted said matching templates according to the number of requests matching each template;

determining cost brackets for each a k value; and

listing said cost brackets in a top k template table.

12. A method of allocating resources as in claim 11, wherein said cost brackets include for each top k list, the number of requests satisfied at minimum cost, above minimum cost and unsatisfied.

13. A method of allocating resources in a cloud environment as in claim 12, wherein templates in the top k list with the lowest associated cost are used for provisioning resources in a cloud environment.

14. A method of allocating cloud environment resources, said method comprising:

matching each resource capacity request included in a server request database with a template specifying allocable cloud resources;

collecting matched templates with at least one match, any templates without matches being discarded;

sorting matched template according to the number of matched requests; and

determining the number of matches for a plurality (k) of matched templates, determined said matches being listed in a top k list.

15. A method of allocating cloud environment resources as in claim 14, wherein each said template specifies an allocable configuration comprising a number of cores, memory space and disk space and matching said each request comprises:

generating a resource match table for listing each matching request and template;

generating a minimum cost list for listing costs associated with each matched template;

selecting a request from said server request database;

identifying a template matching the selected said request;

determining a cost associated with allocable resources specified by the matching said template; and

adding the matched template and request to resource match table and the associated said cost to said minimum cost list.

16. A method of allocating cloud environment resources as in claim 15, wherein matching said each request comprises:

determining if unmatched requests remain in said server request database; and

until no requests remain unselected

returning to selecting a next request and determining a matching template and associated costs for said next request.

17. A method of allocating cloud environment resources as in claim 16, wherein determining top k lists comprises:

generating a table of said matching templates with the number of requests matching allocable resources for each template;

selecting a first k value;

selecting a request from said server request database;

determining cost brackets for the k value;

returning to select a next request until all requests have been selected; and

when all requests are selected

listing said cost brackets in a top k template table.

18. A method of allocating cloud environment resources as in claim 17, wherein said cost brackets include for each top k list, the number of requests satisfied at minimum cost, above minimum cost and unsatisfied and templates in the top k list with the lowest associated cost are used for provisioning resources.

19. A computer program product for allocating cloud environment resources, said computer program product comprising a computer usable medium having computer readable program code stored thereon, said computer readable program code causing a computer executing said code to:

generate a resource match table for listing each matching request and template;

generate a minimum cost list for listing costs associated with each matched template;

select a request from said server request database;

identify a template matching the selected said request;

determine a cost associated with allocable resources specified by the matching said template;

add the matched template and request to resource match table and the associated said cost to said minimum cost list; and

until all requests are selected

return to select a next request and determine a matching template and associated costs for said next request.

20. A computer program product for allocating cloud environment resources as in claim 19, wherein determining top k lists comprises causes said computer executing said code to:

generate a table of said matching templates with the number of requests matching allocable resources for each template;

select a k value;

select a request from said server request database;

determine cost brackets for said k value;

return to select a next request until all requests have been selected;

list said cost brackets in a top k template table when all requests are selected; and

return to select a next k values until all k values are selected.

21. A computer program product for allocating cloud environment resources as in claim 20, wherein said cost brackets include for each top k list, the number of requests satisfied at minimum cost, above minimum cost and unsatisfied, said computer executing said code using said templates in the top k list with the lowest associated cost for provisioning cloud resources.

22. A computer program product for allocating cloud resources, said computer program product comprising a computer usable medium having computer readable program code stored thereon, said computer readable program code comprising:

computer readable program code means for selectively making service provider computer resource capacity available to requesting client devices;

computer readable program code means for storing costs and associated attributes of allocable resources;

computer readable program code means for storing previously received requests for capacity; and

computer readable program code means for selecting and maintaining a selected number of resource templates, each template specifying an allocable resource capacity configuration, each said requesting client device having resources allocated in a cost-aware clustering determined by one of the selected resource templates.

23. A computer program product for allocating cloud resources as in claim 22, further comprising:

computer readable program code means for storing template list options including a complete list option and a top k list option;

computer readable program code means for generating an optimum list of templates to satisfy said previously received requests for each service provider computer; and

computer readable program code means for storing generated template lists.

24. A computer program product for allocating cloud resources as in claim 23, wherein said top k list option indicates a plurality of values for k, said computer readable program code means for generating said optimum list generates template lists for all values of k for said each service provider computer, and selects one of the generated said template lists for said each service provider computer, generated said template lists being stored said top k lists.

25. A computer program product for allocating cloud resources as in claim 23, wherein each said template specifies an allocable configuration comprising a number of cores, memory space and disk space and each selected template list has a minimum cost of said template lists generated for said each service provider computer.