Dynamic processor redistribution between partitions in a computing system

1. An apparatus comprising:

at least one processor;

a memory coupled to the at least one processor;

two or more logical partitions defined on the apparatus, wherein each of the two or more logical partitions owns portions of a shared processing resource; and

a dynamic processor allocation mechanism residing in the memory and executed by the at least one processor, wherein the dynamic processor allocation mechanism reallocates the shared processing resource among the two or more logical partitions and wherein the amount of shared processing resource eligible to be reallocated is calculated on-the-fly by the dynamic processor allocation mechanism prior to the reallocation, the calculation comprising the steps of:

determining a current utilization of the shared processing resource for each of the two or more logical partitions;

determining a current utilization of the shared processing resource for the apparatus as a whole;

constructing, for each of the two or more logical partitions, a bid determined by dividing the current utilization of the partition by the utilization of the apparatus as a whole, wherein any partition having a bid less than 1 is classified as potential donating partition, and any partition having a bid greater than 1 is classified as a potential recipient partition;

calculating, for each potential recipient partition, an amount of resource desired to be received by multiplying the bid by a number of processing units currently assigned to the potential recipient partition, then subtracting the number of processing units currently assigned to the potential recipient partition; and

calculating, for each potential donating partition, an amount of resource available for donation by subtracting the bid from 1, then multiplying the result by the number of processing units currently assigned to the potential donating partition.

2. The apparatus of claim 1 , wherein the execution of the calculation occurs on a periodic, predetermined time interval.

3. The apparatus of claim 1 , wherein a time interval between successive executions of the function is dynamically adjusted.

4. The apparatus of claim 1 , wherein the shared processing resource comprises the at least one processor.

5. The apparatus of claim 1 , wherein a reallocation group is defined to include at least two of the two or more logical partitions, wherein the dynamic processor allocation mechanism is restricted to operate only on logical partitions defined to exist within the allocation group.

6. The apparatus of claim 5 , wherein multiple reallocation groups exist within the apparatus.

7. The apparatus of claim 1 wherein the potential donating partition having the lowest bid is the first partition to donate its resource during the reallocation.

8. The apparatus of claim 1 wherein the potential recipient partition having the highest bid is the first partition to receive resource during the reallocation.

9. The apparatus of claim 1 , wherein the potential recipient partition becomes eligible to receive additional processing resource when its current resource utilization is greater than the current resource utilization for the apparatus as a whole plus a predefined variance.

10. The apparatus of claim 1 , wherein the potential donating partition becomes eligible to donate excess processing resource when its current resource utilization is less than the current resource utilization for the apparatus as a whole minus a predefined variance.

11. A computer-implemented method for managing a shared processing resource in a computer system that includes two or more logical partitions that each own portions of the shared processing resource, the method comprising the steps of:

determining a current utilization of the shared processing resource for each of the two or more logical partitions and a current utilization of the shared processing resource for the computer system as a whole;

constructing, for each of the two or more logical partitions, a bid determined by dividing the current utilization of the partition by the utilization of the apparatus as a whole, wherein any partition having a bid less than 1 is classified as potential donating partition, and any partition having a bid greater than 1 is classified as a potential recipient partition;

calculating, for each potential recipient partition, an amount of resource desired to be received by multiplying the bid by a number of processing units currently assigned to the potential recipient partition, then subtracting the number of processing units currently assigned to the potential recipient partition; and

calculating, for each potential donating partition, an amount of resource available for donation by subtracting the bid from 1, then multiplying the result by the number of processing units currently assigned to the potential donating partition; and

reallocating the shared processing resource among the two or more logical partitions by reassigning at least some of the amount of resource available for donation from the potential donating partitions to the potential recipient partitions.

12. The method of claim 11 , wherein the shared processing resource comprises the at least one processor.

13. A program product comprising:

a dynamic processor allocation mechanism which reallocates a shared processing resource among two or more logical partitions in a computer system, wherein the amount of shared processing resource eligible to be reallocated is calculated on-the-fly by the dynamic processor allocation mechanism prior to the reallocation, the calculation comprising the steps of:

determining a current utilization of the shared processing resource for each of the two or more logical partitions;

determining a current utilization of the shared processing resource for the apparatus as a whole;

constructing, for each of the two or more logical partitions, a bid determined by dividing the current utilization of the partition by the utilization of the apparatus as a whole, wherein any partition having a bid less than 1 is classified as potential donating partition, and any partition having a bid greater than 1 is classified as a potential recipient partition;

calculating, for each potential recipient partition, an amount of resource desired to be received by multiplying the bid by a number of processing units currently assigned to the potential recipient partition, then subtracting the number of processing units currently assigned to the potential recipient partition; and

calculating, for each potential donating partition, an amount of resource available for donation by subtracting the bid from 1, then multiplying the result by the number of processing units currently assigned to the potential donating partition; and computer readable signal bearing recordable media bearing the dynamic processor allocation mechanism.

14. The program product of claim 13 , wherein the shared processing resource comprises at least one processor.