Automated power control policies based on application-specific redundancy characteristics

1. A program product comprising:

a non-transitory computer readable medium having computer readable program code embodied therein, the computer readable program code in said program product being effective in executing the steps of:

(a) determining the workload of each server of a cluster of servers which provide an application service;

(b) determining the topology of the application service wherein said topology determination includes a correspondence between a first component of the application service and the server on which the first component is executed wherein said topology determination includes a first count of the total number of active servers; and

(c) setting the power state of one server of the cluster of servers based upon said workload determination and said topology determination, wherein said setting (c) is to a low power state when said workload determination determines a workload which is below a predetermined threshold, wherein said setting (c) is further a function of a ratio of powered-on inactive servers to the first count and includes a provision for setting a minimum number of powered on inactive servers.

2. The program product of claim 1 wherein said setting (c) is further based upon the extent to which server resources are fully utilized.

3. The program product of claim 1 wherein the low power state is a state selected from the group consisting of: standby, sleep, hibernate, and off states.

4. The program product of claim 1 wherein said setting (c) is to a low power state when said workload determination determines a workload indicative of the elapsed time since the one server was last reset.

5. The program product of claim l wherein said setting (c) is to a powered-on state when said workload determination determines a workload which is above a predetermined threshold.

6. A program product comprising:

a non-transitory computer readable medium having computer readable program code embodied therein, the computer readable program code in said program product being effective in executing the steps of:

(a) determining the workload of each server of a first tier of scalable redundant servers which provide an application service;

(b) determining the topology of the application service wherein said topology determination includes a first count of the total number of active first-tier servers and a correspondence between a first component of the application service and the first-tier server on which the first component is executed;

(c) transitioning a first-tier server between the application service and a free pool of first-tier servers based upon said workload determination; and

(d) controlling the power applied to each of the servers of the free pool of first-tier servers according to a first predetermined power management policy and based upon said topology determination, wherein the first predetermined power management policy applied to the servers of the free pool of first-tier servers maintains a first number of servers in a powered-on inactive state while the remaining servers are set to a low power state, wherein the first predetermined power management policy applied to the servers of the free pool of first-tier servers is a function of a ratio of powered-on inactive servers to the first count and includes a provision for setting a minimum number of powered-on inactive servers.

7. The program product of claim 6 wherein said transitioning is from the application service to the free pool of first-tier servers when said workload determination determines a workload which is below a predetermined threshold.

8. The program product of claim 7 wherein said transitioning is further based upon the extent to which server resources of the transitioning first-tier server are fully utilized.

9. The program product of claim 6 , further comprising:

(e) determining the workload of each server of a second tier of scalable redundant servers which provide the application service wherein responses from the second tier of scalable redundant servers depend on responses from the first tier of scalable redundant servers;

(f) transitioning a second-tier server between the application service and a free pool of second-tier servers; and

wherein said topology determination further includes a second count of the total number of active second-tier servers, and

(g) controlling the power applied to each of the servers of the free pool of second-tier servers according to a second predetermined power management policy and based upon said topology determination;

wherein the second power management policy is a policy selected from the group consisting of the first power management policy and a policy independent of the first power management policy.

10. The program product of claim 6 wherein said transitioning is from the free pool of first-tier servers to the application service when said workload determination determines a workload which is above a predetermined threshold.

11. The program product of claim 6 wherein the low power state is a state selected from the group consisting of: standby, sleep, hibernate, and off states.

12. A method comprising:

(a) determining the workload of each server of a cluster of servers which provide an application service;

(b) determining the topology of the application service wherein said topology determination includes a correspondence between a first component of the application service and the server on which the first component is executed, wherein said topology determination includes a first count of the total number of active servers; and

(c) setting the power state of one server of the cluster of servers based upon said workload determination and said topology determination, wherein said setting (c) is to a low power state when said workload determination determines a workload which is below a predetermined threshold, wherein said setting (c) is further a function of a ratio of powered-on inactive servers to the first count and includes a provision for setting a minimum number of powered-on inactive servers.

13. The method of claim 12 wherein said setting (c) is further based upon the extent to which server resources are fully utilized.

14. The method of claim 12 wherein the low power state is a state selected from the group consisting of: standby, sleep, hibernate, and off states.

15. The method of claim 12 wherein said setting (c) is to a low power state when said workload determination determines a workload indicative of the elapsed time since the one server was last reset.

16. The method of claim 12 wherein said setting (c) is to a powered-on state when said workload determination determines a workload which is above a predetermined threshold.

17. A method comprising:

(a) determining the workload of each server of a first tier of scalable redundant servers which provide an application service;

(b) determining the topology of the application service wherein said topology determination includes a first count of the total number of active first-tier servers and a correspondence between a first component of the application service and the first-tier server on which the first component is executed;

(c) transitioning a first-tier server between the application service and a free pool of first-tier servers based upon said workload determination; and

(d) controlling the power applied to each of the servers of the free pool of first-tier servers according to a first predetermined power management policy and based upon said topology determination , wherein the first predetermined power management policy applied to the servers of the free pool of first-tier servers maintains a first number of servers in a powered-on inactive state while the remaining servers are set to a low power state, wherein the first predetermined power management policy applied to the servers of the free pool of first-tier servers is a function of a ratio of powered-on inactive servers to the first count and includes a provision for setting a minimum number of powered-on inactive servers.

18. The method of claim 17 wherein said transitioning is from the application service to the free pool of first-tier servers when said workload determination determines a workload which is below a predetermined threshold.

19. The method of claim 18 wherein said transitioning is further based upon the extent to which server resources of the transitioning first-tier server are fully utilized.

20. The method of claim 17 , further comprising:

(e) determining the workload of each server of a second tier of scalable redundant servers which provide the application service wherein responses from the second tier of scalable redundant servers depend on responses from the first tier of scalable redundant servers;

(f) transitioning a second-tier server between the application service and a free pool of second-tier servers; and

wherein said topology determination further includes a second count of the total number of active second-tier servers, and

(g) controlling the power applied to each of the servers of the free pool of second-tier servers according to a second predetermined power management policy and based upon said topology determination;

wherein the second power management policy is a policy selected from the group consisting of the first power management policy and a policy independent of the first power management policy.

21. The method of claim 17 wherein said transitioning is from the free pool of first-tier servers to the application service when said workload determination determines a workload which is above a predetermined threshold.

22. The method of claim 17 wherein the low power state is a state selected from the group consisting of: standby, sleep, hibernate, and off states.

23. Apparatus comprising:

a workload monitor which detects the workload of each server of a cluster of servers which provide an application service;

a topology sensor which determines the topology of the application service including a correspondence between a first component of the application service and the server on which the first component is executed, wherein the topology as determined by said topology sensor includes a first count of the total number of active servers; and

a power controller which sets the power state of one server of the cluster of servers based upon the workload as determined by said workload monitor and the topology as determined by said topology sensor, wherein said power controller sets a low power state when said workload monitor determines a workload which is below a predetermined threshold, wherein said power controller sets the power state as a function of a ratio of powered-on inactive servers to the first count and includes a provision for setting a minimum number of powered-on inactive servers.

24. Apparatus of claim 23 wherein said power controller further sets the low power state based upon the extent to which server resources are fully utilized.

25. Apparatus of claim 23 wherein the low power state is a state selected from the group consisting of: standby, sleep, hibernate, and off states.

26. Apparatus of claim 23 wherein said power controller sets the power state to a low power state when said workload monitor determines a workload indicative of the elapsed time since the one server was last reset.

27. Apparatus of claim 23 wherein said power controller sets the power state to a powered-on state when said workload monitor determines a workload which is above a predetermined threshold.

28. Apparatus comprising:

a workload monitor which detects the workload of each server of a first tier of scalable redundant servers which provide an application service;

a topology sensor which determines the topology of the application service including a first count of the total number of active first-tier servers and a correspondence between a first component of the application service and the first-tier server on which the first component is executed;

a move module which transitions a first-tier server between the application service and a free pool of first-tier servers based upon the workload as determined by said workload monitor; and

a power controller which sets the power state of each of the servers of the free pool of first-tier servers according to a first predetermined power management policy and based upon the topology as determined by said topology sensor, wherein the first predetermined power management policy applied to the servers of the free pool of first-tier servers maintains a first number of servers in a powered-on inactive state while the remaining servers are set to a low power state, wherein the first predetermined power management policy applied to the servers of the free pool of first-tier servers is a function of a ratio of powered-on inactive servers to the first count and includes a provision for setting a minimum number of powered-on inactive servers.

29. Apparatus of claim 28 wherein the transition as executed by the move module is from the application service to the free pool of first-tier servers when said workload monitor determines a workload which is below a predetermined threshold.

30. Apparatus of claim 29 wherein the transition as executed by the move module is further based upon the extent to which server resources of the transitioning first-tier server are fully utilized.

31. Apparatus of claim 28

wherein said workload monitor further determines the workload of each server of a second tier of scalable redundant servers which provide the application service wherein responses from the second tier of scalable redundant servers depend on responses from the first tier of scalable redundant servers; and

wherein said move module further transitions a second-tier server between the application service and a free pool of second-tier servers; and

wherein the topology as determined by said topology sensor further includes a second count of the total number of active second-tier servers; and

wherein said power controller further sets the power state of each of the servers of the free pool of second-tier servers according to a second predetermined power management policy and based upon the topology as determined by said topology sensor; and

wherein the second power management policy is a policy selected from the group consisting of the first power management policy and a policy independent of the first power management policy.

32. Apparatus of claim 28 wherein the transition as executed by the move module is from the free pool of first-tier servers to the application service when said workload monitor determines a workload which is above a predetermined threshold.

33. Apparatus of claim 28 wherein the low power state is a state selected from the group consisting of: standby, sleep, hibernate, and off states.