Methods and apparatus for dynamic allocation of servers to a plurality of customers to maximize the revenue of a server farm

• US 20040111509A1
• Filed: 12/10/2002
• Published: 06/10/2004
• Est. Priority Date: 12/10/2002
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Li, Ta-Hsin, Eilam, Tamar, Goldszmidt, German Sergio
• Granted Patent: US 7,783,759 B2
• Time in Patent Office: Days
• US Class Current: 709/224
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/1012 based on compliance of requirements or conditions with available server resources
• CPC Class Codes: H04L 67/1029 using data related to the state of servers by a load balancer

Abstract

A method and structure for dynamic allocation of servers to customers in a server farm which supports a flexible contract structure such that the total revenue of the farm is maximized. The invention creates a resource allocation plan based on a long term forecast for the server farm, taking into account traffic, number of servers, customers'"'"' contracts and revenue optimization algorithms. The plan is then modified as indicated by short term forecasting using currently monitored load metrics to reallocate to maximize revenue for particular time periods.

Claims

1. A method for optimizing server farm resource usage among a plurality of contract customers comprising the steps of:

providing a long term forecast of server resource workload for each contract customer;

building a resource allocation plan for allocation of server resources based on the long term forecast;

controlling the allocation of server resources in accordance with the resource allocation plan;

providing a short term forecast based on monitored workload conditions at said server resources; and

evaluating refinement of the resource allocation plan in accordance with said short term forecast.

2. The method of claim 1 wherein said providing a short term forecast comprises the steps of:

continuously monitoring the use of server resources; and

predicting the load for a next short term time slot.

3. The method of claim 1 wherein said evaluating refinement comprises comparing the short term forecast to the long term forecast.

4. The method of claim 3 further comprising refining the resource allocation plan when the short term forecast differs from the long term forecast by more than a pre-established threshold.

5. The method of claim 1 wherein said evaluating is done periodically according to a preset schedule.

6. The method of claim 3 further comprising generating a short term forecast alert when said comparing indicates that said short term forecast differs from the long term forecast by more than a pre-established threshold.

7. The method of claim 1 wherein said refining is done in response to a short term forecast alert.

8. The method of claim 7 further comprising refining said resource allocation plan by reallocating server resources.

9. The method of claim 1 wherein said building of said resource allocation plan further comprises optimizing revenues for said server farm while addressing server resource workload in said long term forecast.

10. The method of claim 9 wherein said step of optimizing revenues comprises evaluating customer contract terms.

11. A system for optimizing server farm resource usage among a plurality of contract customers comprising:

a long term forecasting component for generating a long term forecast of server resource workload;

a resource allocation planning component for constructing a resource allocation plan for allocation of server resources based on the long term forecast;

a short term forecasting component for generating a short term forecast of server resource workload; and

a resource manager for controlling the allocation of server resources in accordance with the resource allocation plan and for evaluating refinement of said resource allocation plan based on the short term forecast.

12. The system of claim 11 further comprising a workload monitoring component for monitoring current resource request information and for providing said information to the short term forecasting component.

13. The system of claim 12 wherein said short term forecasting component predicts the load for the next short term time slot.

14. The system of claim 11 further comprising a comparison component for comparing the short term forecast to the long term forecast.

15. The system of claim 14 wherein the resource manager reallocates server resources based on output from said comparison component.

16. The system of claim 11 further comprising a timing component for periodically causing said short term forecasting component to generate a short term forecast.

17. The system of claim 14 wherein said comparison component additionally comprises at least one storage location for maintaining a preset threshold and wherein said comparison component generates a short term forecast alert when said short term forecast differs from said long term forecast by more than said preset threshold.

18. The system of claim 11 wherein said system additionally comprises at least one storage location for maintaining a preset threshold and wherein said short term forecast component generates a short term forecast alert when said short term forecast differs from said long term forecast by more than said preset threshold.

19. The system of claim 11 wherein said resource allocation planning component applies a revenue optimization function for allocating server resources based on customer contracts.

20. A program storage device readable by machine, tangibly embodying a program of instructions for said machine to execute a method for optimizing server farm resource usage among a plurality of contract customers, said method comprising the steps of:

providing a long term forecast of server resource workload for each contract customer;

building a resource allocation plan for allocation of server resources based on the long term forecast;

controlling the allocation of server resources in accordance with the resource allocation plan;

providing a short term forecast based on monitored workload conditions at said server resources; and

evaluating refinement of the resource allocation plan in accordance with said short term forecast.