Computer resource utilization modeling for multiple workloads

1. A method, comprising:

making, via a computer, utilization measurements of at least one computer resource during time intervals for a plurality of workloads;

grouping, via the computer, the computer resource utilization measurements into groups;

determining, via the computer, parameters representing a probability distribution of the computer resource utilization measurements for each of the groups;

generating, via the computer, from the determined parameters; a baseline as a set of the probability distributions across the time intervals for each of the plurality of workloads; and

generating, via the computer, a combination of the baselines as a combined baseline of probability distributions across the time intervals for a combination of the plurality of workloads.

2. The method of claim 1 , wherein the determined parameters for each of the groups include at least one of a location parameter, a scale parameter, or a shape parameter.

3. The method of claim 1 , wherein generating the combination of the baselines includes summing the combination of the baselines.

4. The method of claim 1 , wherein the determined parameters for each of the groups include a mean and one selected from a group comprising a percentile and a standard deviation.

5. The method of claim 1 , further comprising:

scaling the determined parameters for the groups corresponding to at least one of the plurality of workloads.

6. An apparatus to model computer resource utilization, comprising:

a memory to store computer resource utilization data, the computer resource utilization data having a plurality of workload data sets for a respective plurality of workloads, each of the plurality of workload data sets including computer resource utilization measurements made across a plurality of time intervals; and

a processor to access the computer resource utilization data in the memory, to process the computer resource utilization measurements in the plurality of workload data sets to generate parameters representing per-workload baselines of probability distributions across the plurality of time intervals for the plurality of workload data sets, and to process the parameters to generate a combined baseline of probability distributions for a combination of the plurality of workloads from the per-workload baselines of probability distributions.

7. The apparatus of claim 6 , wherein the parameters include at least one of a location parameter, a scale parameter, or a shape parameter for the computer resource utilization measurements for each of the plurality of time intervals for each of the plurality of workload data sets.

8. The apparatus of claim 6 , wherein the processor scales the parameters for at least one of the workload data sets.

9. The apparatus of claim 6 , wherein the processor processes the combined baseline of probability distributions for the combination of the plurality of workloads to generate at least one metric.

10. The apparatus of claim 9 , wherein the at least one metric includes an estimate of a mean computer resource utilization by the combination of the plurality of workloads.

11. The apparatus of claim 9 , wherein the at least one metric includes an estimate of peak computer resource utilization by the combination of the plurality of workloads in at least one of the plurality of time intervals.

12. The apparatus of claim 9 , wherein the at least one metric includes an estimate of computer resource utilization expected to be exceeded by the combination of the plurality of workloads a defined percentage of time within the plurality of time intervals.

13. A method, comprising:

obtaining, via a computer resource, utilization data of the computer resource for a plurality of workloads over a plurality of time intervals;

determining, via the computer resource, probability distribution parameters from the computer resource utilization data for each of the plurality of time intervals;

generating, via the computer resource, an individual baseline across the plurality of time intervals for each of the plurality of workloads from the probability distribution parameters;

summing, via the computer resource, the individual baselines for the plurality of workloads into a combined baseline across the plurality of time intervals for a combination of the plurality of workloads; and

determining, via the computer resource, at least one metric from the combined baseline.

14. The method of claim 13 , wherein the probability distribution parameters include at least one of location parameters, scale parameters, and shape parameters.

15. The method of claim 13 , wherein the probability distribution parameters for each of the groups include a mean and a percentile.

16. The method of claim 13 , wherein the probability distribution parameters for each of the groups include a mean and a standard deviation.

17. The method of claim 13 , further comprising: scaling, via the computer resource, the probability distribution parameters for at least one of the plurality of workloads.

18. The method of claim 13 , wherein the at least one metric includes an estimate of a mean computer resource utilization by the combination of the plurality of workloads.

19. The method of claim 13 , wherein the at least one metric includes an estimate of peak computer resource utilization by the combination of the plurality of workloads in at least one of the plurality of time intervals.

20. The method of claim 13 , wherein the at least one metric includes an estimate of computer resource utilization expected to be exceeded by the combination of the plurality of workloads a defined percentage of time within the plurality of time intervals.