System and method for fair and economical resource partitioning using virtual hypervisor

• US 8,433,802 B2
• Filed: 01/26/2010
• Issued: 04/30/2013
• Est. Priority Date: 01/26/2010
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Head, Michael R., Kochut, Andrzej, Schulz, Charles O., Shaikh, Hidayatullah H.
• Primary Examiner(s): Dalencourt, Yves
• Assistant Examiner(s): Kim, Hee Soo
• Application Number:
  US12/693,967
• Publication Number:
  US 20110185064A1
• Time in Patent Office: 1,190 Days
• Field of Search: 709/219, 709/226, 709/223, 718/1
• US Class Current: 709/226
• CPC Class Codes: G06F 2209/504 Resource capping
• CPC Class Codes: G06F 9/45533 Hypervisors; Virtual machine monitors
• CPC Class Codes: G06F 9/5077 Logical partitioning of resources; Management or configuration of virtualized resources specific details on emulation or internal functioning of virtual machines G06F9/455
• CPC Class Codes: G06F 9/5088 involving task migration
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A system and method for allocating resources in a cloud environment includes determining permitted usage of virtual machines and partitioning resources between network servers in accordance with a virtual hypervisor generated in accordance with an abstraction layer configured as an interface between a solution manager and an interface to a cloud network. Resource usage limits are determined for each virtual machine associated with the virtual hypervisor, and the servers are analyzed through the virtual hypervisors to determine if the virtual machines need to be migrated. If reallocation is needed, virtual machine migration requests are issued to migrate virtual machines into a new configuration at the virtual hypervisor abstraction level. The servers are reanalyzed to determine if migration of the new configuration is needed. Shares are computed to enforce balance requirements, and virtual machine shares and limits are adjusted for resources according to the computed shares.

Claims

1. A method for allocating resources in a cloud environment, comprising:

determining permitted usage for each of a plurality of virtual machines;

partitioning resources and virtual machines between network servers in accordance with a virtual hypervisor manager generated in accordance with a virtual hypervisor abstraction layer configured as an interface between a solution manager and an interface to a cloud network; and

specifying resource allocation in the cloud environment using a virtual hypervisor, allowing the solution manager to dynamically control resource allocation decisions while maintaining a cloud manager'"'"'s role as an ultimate physical resource manager.

2. The method as recited in claim 1, further comprising:

determining resource usage limits for each virtual machine associated with the virtual hypervisor;

analyzing the servers through the virtual hypervisor to determine if the virtual machines need to be migrated to keep the servers within operating utilization range; and

if reallocation of resources is needed, issuing virtual machine migration requests to migrate virtual machines into a new configuration at the virtual hypervisor abstraction level.

3. The method as recited in claim 2, further comprising:

reanalyzing the servers to determine if the virtual machines need to be further migrated in the new configuration;

computing shares to be set on resource and virtual machines to enforce balance requirements; and

adjusting virtual machine shares and limits for all tracked resources according to computed shares.

4. The method as recited in claim 3, further comprising executing the method for analysis and resource adjustment iteratively employing a tunable step size.

5. The method as recited in claim 2, wherein determining resource usage limits includes determining a capacity of the virtual hypervisor.

6. The method as recited in claim 1, wherein determining permitted usage includes determining permitted usage based upon service level agreements.

7. The method as recited in claim 1, wherein the virtual hypervisor abstraction layer generates an application programming interface to manage the resources and virtual machines.

8. The method as recited in claim 1, wherein the cloud network includes a cloud manager for mapping virtual machines to physical servers in accordance with constraints defined by at least one virtual hypervisor, wherein the constraints defined includes a minimum guaranteed rate of growth of a resource for a virtual machine.

9. The method as recited in claim 1, wherein the virtual hypervisor is requested from the cloud network.

10. The method as recited in claim 1, wherein the virtual hypervisor includes abstract groupings of resources which are managed by a cloud manager of the cloud network with respect to a service level agreement between the cloud manager of the cloud network and the solution manager.

11. The method as recited in claim 1, wherein deployment and management of the virtual machines on top of the virtual hypervisor is handled by the solution manager.

12. A non-transitory computer readable storage medium comprising a computer readable program for allocating resources in a cloud environment, wherein the computer readable program when executed on a computer causes the computer to perform the steps of:

determining permitted usage for each of a plurality of virtual machines;

partitioning resources and virtual machines between network servers in accordance with a virtual hypervisor manager generated in accordance with a virtual hypervisor abstraction layer configured as an interface between a solution manager and an interface to a cloud network; and

specifying resource allocation in the cloud environment using a virtual hypervisor, allowing the solution manager to dynamically control resource allocation decisions while maintaining a cloud manager'"'"'s role as an ultimate physical resource manager.

13. The computer readable storage medium as recited in claim 12, further comprising:

determining resource usage limits for each virtual machine associated with the virtual hypervisor;

analyzing the servers through the virtual hypervisor to determine if the virtual machines need to be migrated to keep the servers within operating utilization range; and

if reallocation of resources is needed, issuing virtual machine migration requests to migrate virtual machines into a new configuration at the virtual hypervisor abstraction level.

14. The computer readable storage medium as recited in claim 13, further comprising:

reanalyzing the servers to determine if the virtual machines need to be further migrated in the new configuration;

computing shares to be set on resource and virtual machines to enforce balance requirements; and

adjusting virtual machine shares and limits for all tracked resources according to computed shares.

15. The computer readable storage medium as recited in claim 14, further comprising executing the method for analysis and resource adjustment iteratively employing a tunable step size.

16. The computer readable storage medium as recited in claim 13, wherein determining resource usage limits includes determining a capacity of the virtual hypervisor.

17. The computer readable storage medium as recited in claim 12, wherein determining permitted usage includes determining permitted usage based upon service level agreements.

18. The computer readable storage medium as recited in claim 12, wherein the virtual hypervisor abstraction layer generates an application programming interface to manage the resources and virtual machines.

19. The computer readable storage medium as recited in claim 12, wherein the cloud network includes a cloud manager for mapping virtual machines to physical servers in accordance with constraints defined by at least one virtual hypervisor, wherein the constraints defined include a minimum guaranteed rate of growth of a resource for a virtual machine.

20. The computer readable storage medium as recited in claim 12, wherein the virtual hypervisor is requested from the cloud network.

21. The computer readable storage medium as recited in claim 12, wherein the virtual hypervisor includes abstract groupings of resources which are managed by a cloud manager of the cloud network with respect to a service level agreement between the cloud manager of the cloud network and the solution manager.

22. The computer readable storage medium as recited in claim 12, wherein deployment and management of the virtual machines on top of the virtual hypervisor is handled by the solution manager.

23. A system for allocating resources in a cloud environment, comprising:

a solution manager configured to deploy and manage virtual machines;

a cloud network managed by a cloud manager and including at least on physical server configured to employ the virtual machines;

at least one virtual hypervisor including abstract groupings of resources which are managed by the cloud manager with respect to a service level agreement between the cloud manager of the cloud network and the solution manager, wherein the cloud manager maps the virtual machines to the at least one physical server in accordance with constraints defined by the at least one virtual hypervisor and the virtual machines handled by the solution manager are abstracted on top of the at least one virtual hypervisor such that resources are efficiently and fairly allocated within the cloud network; and

the solution manager configured to dynamically control resource allocation decisions within the cloud environment while maintaining a cloud manager'"'"'s role as an ultimate physical resource manager.

24. The system as recited in claim 23, further comprising a virtual hypervisor application programming interface configured to manage the resources and virtual machines by providing an interface between a cloud application programming interface and the solution manager.

25. The system as recited in claim 24, further comprising physical hypervisors associated with physical servers wherein the virtual hypervisor application programming interface provides a user with application programming interfaces of the physical hypervisor without exposing the details of particular hypervisor software or upon which physical server the hypervisor runs.