Data center power scenario simulation
In disclosed techniques, simulations are performed to determine data center performance under certain conditions. The simulations are dynamic and allow for changes in power demand due to temporal data center activities. In order to accommodate predicted and unpredicted fluctuations in power demand of a data center, one or more power caches are configured to supply additional power during periods of peak demand. Power caches provide supplemental power during periods of peak demand. The simulations are used for a variety of purposes, including determining the effects of power caches going offline under various conditions. Disclosed techniques can simulate the cycling of a power cache and can determine if additional configuration changes to the data center are warranted to maintain optimal health of the power caches. Thus, power scenario simulation of a data center can provide information vital to efficient operation of the data center.
1. A computer-implemented method for power analysis comprising:
obtaining, on a first computing device, a description of a first set of power loads for a data center;
obtaining, on the first computing device, a description of a first set of power supplies coupled to the first set of power loads wherein the first set of power supplies is further coupled to a first set of power caches; and
simulating, on a second computing device, a dynamic power scenario with the first set of power loads, the first set of power supplies, and the first set of power caches wherein the simulating includes evaluating the first set of power loads based on applications running on a load platform within the first set of power loads, and wherein the simulating further includes evaluating an estimated completion time for a task associated with at least one of the applications, based on a number of processing cores allocated for the task, wherein the simulating includes evaluating the first set of power loads using a first randomized pattern of component usage and performance for the first set of power loads, wherein the first randomized pattern of component usage and performance includes:
a power supply randomization range for a power supply, and wherein the power supply randomization range has a zero minimum value and a maximum output based on a maximum operating condition of the power supply;
one or more of the power supplies coming offline at various random times;
one or more of the power loads being activated at random times;
power consumption increasing at random times; and
power caches failing at random times.
2. The method of claim 1 wherein the first set of power loads is distributed across a plurality of data racks.
3. The method of claim 1 wherein the first set of power caches provides power when power requirements from the first set of power loads exceed capabilities of the first set of power supplies.
4. The method of claim 3 wherein the capabilities of the first set of power supplies are defined by parameter policies.
5. The method of claim 4 further comprising validating a set of power policies based on the simulating wherein the power policies determine how power is allocated among various devices within the data center.
6. The method of claim 5 wherein the validating further comprises validation of a power configuration topology within the data center.
7. The method of claim 6 wherein the simulating identifies that the data center is operating outside a preferred efficiency range and updates the power configuration topology to improve efficiency.
8. The method of claim 6 further comprising changing topology of the data center as part of the validating and re-simulating the dynamic power scenario.
9. The method of claim 8 wherein the changing topology includes removing part of the first set of power supplies.
10. The method of claim 8 wherein the changing topology includes removing part of the first set of power caches.
11. The method of claim 8 wherein the changing topology includes removing part of the first set of power loads.
12. The method of claim 8 wherein the re-simulating includes evaluation of a conjecture scenario for power consumption.
13. The method of claim 1 wherein the simulating of the dynamic power scenario covers a period of time in the data center.
14. The method of claim 13 wherein the simulating analyzes a previous power experience by the data center.
15. The method of claim 14 wherein the previous power experience includes a failure power scenario.
16. The method of claim 15 wherein the simulating includes a resolution to the previous power experience that includes the failure for the dynamic power scenario.
17. The method of claim 1 wherein the simulating analyzes a failure for the dynamic power scenario.
18. The method of claim 17 wherein the failure includes a capacity, for the first set of power supplies and the first set of power caches, being exceed by the first set of power loads.
19. The method of claim 1 wherein the simulating further includes evaluating the first set of power loads with a second randomized pattern of component usage for the first set of power loads.
20. The method of claim 19 further comprising selecting one of the first randomized patterns of component usage and the second randomized pattern of component usage, based on a selection criterion, for operation of the first set of power loads.
21. The method of claim 1 wherein the simulating includes validating a data center power configuration.
22. The method of claim 1 wherein the simulating includes providing recommendations for changing topology of the data center.
23. The method of claim 1 wherein the first computing device and the second computing device are a common computing device.
24. The method of claim 1 further comprising providing information on the power scenario and the simulating to a cloud analysis facility.
25. The method of claim 1 further comprising augmenting the simulating with information from a cloud analysis facility.
26. The method of claim 1 wherein the simulating further includes evaluating, for each processing core from the number of processing cores, a core type, wherein the core type includes a single core processing unit and a multiple core processing unit.
27. A computer program product embodied in a non-transitory computer readable medium for power analysis, the computer program product comprising code which causes one or more processors to perform operations of:
obtaining, on a first computing device, a description of a first set of power supplies coupled to the first set of power loads wherein the first set of power supplies is further coupled to a first set of power caches; and
simulating, on a second computing device, a dynamic power scenario with the first set of power loads, the first set of power supplies, and the first set of power caches wherein the simulating includes evaluating the first set of power loads based on applications running on a load platform within the first set of power loads, and wherein the simulating further includes evaluating an estimated completion time for a task associated with at least one of the applications, based on a number of processing cores allocated for the task, wherein the simulating includes evaluating the first set of power loads using a first randomized pattern of component usage and performance for the first set of power loads, wherein the first randomized pattern of component usage and performance includes:
a power supply randomization range for a power supply, and wherein the power supply randomization range has a zero minimum value and a maximum output based on a maximum operating condition of the power supply; and
power consumption increasing at random times.
28. A computer system for power analysis comprising:
a memory which stores instructions;
one or more processors attached to the memory wherein the one or more processors, when executing the instructions which are stored, are configured to:
obtain a description of a first set of power loads for a data center;
obtain a description of a first set of power supplies coupled to the first set of power loads, wherein the first set of power supplies is further coupled to a first set of power caches;
simulate a dynamic power scenario with the first set of power loads, the first set of power supplies, and the first set of power caches; and
evaluate the first set of power loads based on applications running on a load platform within the first set of power loads, and wherein simulation further includes evaluating an estimated completion time for a task associated with at least one of the applications, based on a number of processing cores allocated for the task, wherein the simulating includes evaluating the first set of power loads using a first randomized pattern of component usage and performance for the first set of power loads, wherein the first randomized pattern of component usage and performance includes:
a power supply randomization range for a power supply, and wherein the power supply randomization range has a zero minimum value and a maximum output based on a maximum operating condition of the power supply; and
power cache failures.