On-demand application resource allocation through dynamic reconfiguration of application cluster size and placement

• US 20060095917A1
• Filed: 11/01/2004
• Published: 05/04/2006
• Est. Priority Date: 11/01/2004
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Steinder, Malgorzata, Pacifici, Giovanni, Tantawi, Asser N., Martin, Brian K., Black-Ziegelbein, Elizabeth A., Spreitzer, Michael, Cuomo, Gennaro A.
• Granted Patent: US 7,788,671 B2
• Time in Patent Office: Days
• US Class Current: 718/104
• CPC Class Codes: G06F 9/5061 Partitioning or combining of resources
• CPC Class Codes: G06F 9/5066 Algorithms for mapping a plurality of inter-dependent sub-tasks onto a plurality of physical CPUs mappping at compile time, see G06F8/451
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system

Abstract

A method, system and apparatus for on-demand application resource allocation. In accordance with the method of the invention, an anticipated workload can be compared to a measured capacity for an application cluster in one or more server computing nodes in a server farm. If the measured capacity warrants a re-configuration of the application clusters, a new placement can be computed for application server instances in individual ones of the server computing nodes. Subsequently, the new placement can be applied to the server computing nodes in the server farm. In this regard, the applying step can include starting and stopping selected ones of the application server instances in different ones of the server computing nodes to effectuate the new placement.

Claims

1. A method for on-demand application resource allocation by dynamically reconfiguring the size and placement of application clusters, the method comprising the steps of:

comparing an anticipated workload to a measured capacity for the application clusters in the server computing nodes in a server farm; and

,

if said measured capacity warrants a re-configuration of said application clusters, computing a new placement for application server instances in individual ones of said server computing nodes, and applying said new placement to said server computing nodes in said server farm.

2. The method of claim 1, wherein said comparing step comprises the steps of:

predicting a load-dependent demand;

additionally predicting a load-independent demand;

measuring a load-dependent capacity;

additionally measuring a load-independent capacity; and

,

comparing said demands to said capacity to determine whether a re-configuration of said server computing nodes is warranted.

3. The method of claim 1, further comprising the step of repeating the comparing, computing and applying steps periodically for different anticipated workloads.

4. The method of claim 1, wherein said applying step comprises the step of starting and stopping selected ones of said application server instances in different ones of said server computing nodes to effectuate said new placement.

5. The method of claim 1, further comprising the step of limiting said computing step based upon the terms of a placement policy.

6. The method of claim 5, wherein said limiting step comprises the step of limiting said computing step to ensure that a placement for a manually designated cluster is not changed.

7. The method of claim 5, wherein said limiting step comprises the step of limiting said computing step to ensure that a total number of application server instances in an application cluster is at least a specified minimum value.

8. The method of claim 5, wherein said limiting step comprises the step of limiting said computing step to ensure that a total number of application server instances in an application cluster is no more than a specified maximum value.

9. The method of claim 1, further comprising the step of limiting said applying step based upon the terms of a placement policy.

10. The method of claim 9, wherein said limiting step comprises the step of limiting said applying step to ensure the application of no more than one change to any of said server computing nodes at any one time.

11. The method of claim 9, wherein said limiting step comprises the step of limiting said applying step to ensure that a total number of application server instances in an application cluster is at least a specified minimum value.

12. The method of claim 9, wherein said limiting step comprises, the step of limiting said applying step to obey load-independent capacity constraints known for each of said server computing nodes.

21. The machine readable storage of claim 12, wherein said applying step comprises the step of starting and stopping selected ones of said application server instances in different ones of said server computing nodes to effectuate said new placement.

13. An on-demand application resource allocation system comprising:

a plurality of server computing nodes configured to host a plurality of application clusters comprising a plurality of application server instances;

at least one request router coupled to said server computing nodes; and

,

a placement executor communicatively linked to said application server instances and programmed to apply a computed placement by starting and stopping selected ones of said application server instances.

14. The system of claim 13, further comprising a placement controller programmed to compute said desired placement.

15. The system of claim 14, further comprising each of an application workload predictor and an application profiler configured to produce a workload prediction and a set of application resource requirements utilized by said placement controller to compute said desired placement.

16. The system of claim 14, wherein said placement controller further comprises logic configured to limit a computation of said desired placement based upon a specified placement policy.

17. The system of claim 13, wherein said placement executor further comprises logic configured to limit an application of said desired placement based upon a specified placement policy.

18. A machine readable storage having stored thereon a computer program for on-demand application resource allocation, the computer program comprising a routine set of instructions which when executed by a machine cause the machine to perform the steps of:

comparing an anticipated workload to a measured capacity for the application clusters in the server computing nodes in a server farm; and

,

if said measured capacity warrants a re-configuration of said application clusters, computing a new placement for application server instances in individual ones of said server computing nodes, and applying said new placement to said server computing nodes in said server farm.

19. The machine readable storage of claim 18, wherein said comparing step comprises the steps of:

predicting a load-dependent demand;

additionally predicting a load-independent demand;

measuring a load-dependent capacity;

additionally measuring a load-independent capacity; and

,

comparing said demands to said capacity to determine whether a re-configuration of said server computing nodes is warranted.

20. The machine readable storage of claim 18, further comprising an additional set of instructions for causing the machine to perform the step of repeating the comparing, computing and applying steps periodically for different anticipated workloads.

22. The machine readable storage of claim 18, further comprising an additional set of instructions for causing the machine to perform the step of limiting said computing step based upon the terms of a placement policy.

23. The machine readable storage of claim 22, wherein said limiting step comprises the step of limiting said computing step to ensure that a placement for a manually designated cluster is not changed.

24. The machine readable storage of claim 22, wherein said limiting step comprises the step of limiting said computing step to ensure that a total number of application server instances in an application cluster is at least a specified minimum value.

25. The machine readable storage of claim 22, wherein said limiting step comprises the step of limiting said computing step to ensure that a total number of application server instances in an application cluster is no more than a specified maximum value.

26. The machine readable storage of claim 18, further comprising an additional set of instructions for causing the machine to perform the step of limiting said applying step based upon the terms of a placement policy.

27. The machine readable storage of claim 26, wherein said limiting step comprises the step of limiting said applying step to ensure the application of no more than one change to any of said server computing nodes at any one time.

28. The machine readable storage of claim 26, wherein said limiting step comprises the step of limiting said applying step to ensure that a total number of application server instances in an application cluster is at least a specified minimum value.

29. The machine readable storage of claim 26, wherein said limiting step comprises the step of limiting said applying step to obey load-independent capacity constraints known for each of said server computing nodes.