Resource allocation in data centers using models

• US 20040073673A1
• Filed: 10/10/2002
• Published: 04/15/2004
• Est. Priority Date: 10/10/2002
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Zhu, Xiaoyun, Santos, Cipriano A., Crowder, Harlan
• Granted Patent: US 7,743,127 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: H04L 69/329 in the application layer [OSI layer 7]
• CPC Class Codes: H04L 9/40 Network security protocols

Abstract

A method for resource allocation in a data center is disclosed which employs modeling techniques that enable resource allocation in large-scale data centers. A method for resource allocation according to the present techniques includes providing a model for resource allocation which is adapted to a resource topology of a data center and an architecture of an application to be deployed in the data center. A set of resources in the data center are allocated to the application using the model in response to a set of resource requirements of the application and a set of attributes of the resources in the data center such that communication delay among the resources allocated to the application is minimized. Techniques are disclosed for decomposing the model into a set of more tractable programming models which may be solved using available mathematical programming solvers.

Claims

1. A method for resource allocation in a data center, comprising the steps of:

providing a model for resource allocation which is adapted to a topology of the data center and an architecture of an application to be deployed in the data center;

allocating a set of resources in the data center to the application using the model in response to a set of resource requirements of the application and a set of attributes of the resources in the data center.

2. The method of claim 1, wherein the step of providing a model for resource allocation comprises the step of providing a mathematical model for resource allocation which is adapted to the topology of the data center and the architecture of the application.

3. The method of claim 1, wherein the step of allocating comprises the step of allocating the resources using the model such that communication delay among the resources allocated to the application is minimized.

4. The method of claim 1, wherein the step of allocating a set of resources in the data center to the application using the model comprises the steps of:

determining an approximate solution to the model using a substitute model that does not include one or more constraints associated with the resource requirements and the attributes;

determining a solution to a relaxation of the model which is based on feasibility of resource allocation with respect to the constraints not included in the substitute model;

identifying the resources allocated to the application using a model for rounding the solution.

5. The method of claim 4, wherein the step of determining a solution to a relaxation of the model comprises the step of determining a solution to a relaxation of the model which is based on feasibility of resource allocation with respect to the constraints not included in the substitute model and which uses the approximate solution as an initial solution.

6. The method of claim 4, wherein the step of determining an approximate solution to the model using a substitute model that does not include one or more constraints comprises the step of determining an approximate solution to the model using a substitute model that does not include one or more of a set of bandwidth constraints.

7. The method of claim 1, wherein the step of providing a model for resource allocation which is adapted to a topology of the data center comprises the step of providing a model for resource allocation which is adapted to a hierarchical tree topology.

8. The method of claim 1, wherein the step of providing a model for resource allocation which is adapted to an architecture of an application comprises the step of providing a model for resource allocation which is adapted to a multi-layer architecture.

9. A resource deployment system, comprising:

model for resource allocation which is adapted to a topology of a data center and an architecture of an application to be deployed in the data center;

resource allocator that identifies a set of resources in the data center that are to be allocated to the application using the model in response to a set of resource requirements of the application and a set of attributes of the resources in the data center.

10. The resource deployment system of claim 9, wherein the model is a mathematical model.

11. The resource deployment system of claim 9, wherein the model is adapted to minimize communication delay among the resources allocated to the application.

12. The resource deployment system of claim 9, wherein the resource allocator uses the model by determining an approximate solution to the model using a substitute model that does not include one or more constraints associated with the resource requirements and the attributes, determining a solution to a relaxation of the model which is based on feasibility of resource allocation with respect to the constraints not included in the substitute model, and then rounding the solution.

13. The resource deployment system of claim 12, wherein the resource allocator uses the approximate solution as an initial solution to the relaxation of the model.

14. The resource deployment system of claim 12, wherein the substitute model does not include one or more of a set of bandwidth constraints.

15. The resource deployment system of claim 9, wherein the topology of the data center is a hierarchical tree topology.

16. The resource deployment system of claim 9, wherein the architecture of the application is a multi-layer architecture.

17. A computer-readable storage media that contains a program that when executed by a computer allocates resources in a data center by performing the step of allocating a set of resources in the data center to the application using a model for resource allocation in response to a set of resource requirements of the application and a set of attributes of the resources in the data center wherein the model is adapted to a topology of the data center and an architecture of an application to be deployed in the data center;

18. The computer-readable storage media of claim 17, wherein the model for resource allocation is a mathematical model for resource allocation which is adapted to the topology of the data center and the architecture of the application.

19. The computer-readable storage media of claim 17, wherein the step of allocating comprises the step of allocating the resources using the model such that communication delay among the resources allocated to the application is minimized.

20. The computer-readable storage media of claim 17, wherein the step of allocating a set of resources in the data center to the application using the model comprises the steps of:

determining an approximate solution to the model using a substitute model that does not include one or more constraints associated with the resource requirements and the attributes;

determining a solution to a relaxation of the model which is based on feasibility of resource allocation with respect to the constraints not included in the substitute model;

identifying the resources allocated to the application using a model for rounding the solution.

21. The computer-readable storage media of claim 20, wherein the step of determining a solution to a relaxation of the model comprises the step of determining a solution to a relaxation of the model which is based on feasibility of resource allocation with respect to the constraints not included in the substitute model and which uses the approximate solution as an initial solution.

22. The computer-readable storage media of claim 20, wherein the step of determining an approximate solution to the model using a substitute model that does not include one or more constraints comprises the step of determining an approximate solution to the model using a substitute model that does not include one or more of a set of bandwidth constraints.

23. The computer-readable storage media of claim 17, wherein the model for resource allocation is adapted to a hierarchical tree topology of the data center.

24. The computer-readable storage media of claim 17, wherein the model for resource allocation is adapted to a multi-layer architecture of the application.