Forecast for demand of energy

Patent metadata

Field	Value
Publication number	US-2016187395-A1
Application number	US-201414582988-A
Country	US
Kind code	A1
Filing date	Dec 24, 2014
Priority date	Dec 24, 2014
Publication date	Jun 30, 2016
Grant date	—

Abstract

A system and method for forecasting power consumption at a facility, the facility having a system of compute units for executing jobs of computing. The forecast of power includes forecasting sequence of jobs execution on a system of the nodes over time, estimating power for the jobs of the system, and developing a system-level power forecast.

First claim

What is claimed is:

1 . A distributed computing facility comprising:
nodes configured to execute a job of distributed computing; and memory storing code executable by the nodes or a processor to forecast power consumption by the distributed computing facility, comprising to:
forecast job sequence of jobs of distributed computing on a system of the nodes over time;
estimate power for the jobs on the system; and
develop a system-level power forecast of the system.

2 . The facility of claim 1 , wherein the system-level power forecast comprises a predicted power profile correlative with the job sequence and the power for the jobs.

3 . The facility of claim 1 , wherein the code is executable to activate or increase local energy generation in response to a forecast of power consumption exceeding a specified demand.

4 . The facility of claim 1 , wherein to forecast power consumption comprises to develop a forecast of total actual demand for the distributed computing facility.

5 . The facility of claim 4 , wherein to develop the forecast of total actual demand comprises considering local power generation at the distributed computing facility.

6 . The facility of claim 4 , wherein the total actual demand comprises the power consumption minus an amount of local power generated at the distributed computing facility.

7 . The facility of claim 1 , wherein to forecast power consumption comprises to forecast power consumption in response to a job event or an energy event, or a combination thereof.

8 . The facility of claim 7 , wherein the job event comprise submission of a job for execution to the system, completion of a job on the system, suspension of a job on the system, or any combination thereof.

9 . The facility of claim 7 , wherein the energy event comprises a change in status or capacity of local generation of energy, change in capacity of local storage of energy, or change in pricing of energy beyond predefined thresholds, or any combination thereof.

10 . The facility of claim 1 , wherein the code is executable to adjust total actual demand of power by the distributed computing facility by adjusting use of local energy generation or energy storage, or both, at the distributed computing facility.

11 . The facility of claim 1 , wherein the code is executable to adjust use of local energy generation or energy storage, or both, at the distributed computing facility in response to changing demand of power or changing pricing of energy, or both.

12 . The facility of claim 1 , wherein the code is executable to adjust total actual demand of power by the distributed computing facility in response to an incentive program by a utility provider or in response to spot pricing of energy, or a combination thereof.

13 . The facility of claim 1 , wherein the distributed computing facility comprises a high performance computing (HPC) facility or a Big Data analytics facility, or a combination thereof.

14 . A method of operating a distributed computing facility, comprising:
executing jobs of distributed computing on systems of nodes of the distributed computing facility; forecasting, via a processor, power consumption of the distributed computing facility, wherein forecasting the power consumption comprises:
forecasting, via the processor, a sequence of jobs execution on a system of the nodes over time;
estimating, via the processor, power for the jobs of the system; and
developing, via the processor, a system-level power forecast of the system.

15 . The method of claim 14 , wherein the system-level power forecast comprises a predicted power profile correlative with the sequence of job execution and the power for the jobs.

16 . The method of claim 14 , comprising activating or increasing local energy generation at the distributed computing facility in response to a forecast of the power consumption exceeding a specified demand.

17 . The method of claim 14 , wherein forecasting power consumption comprises forecasting total actual demand to a utility provider for the distributed computing facility.

18 . The method of claim 17 , wherein forecasting total actual demand comprises considering local power generation at the distributed computing facility.

19 . The method of claim 14 , wherein forecasting power consumption comprises forecasting power consumption in response to a job event or an energy event.

20 . The method of claim 14 , comprising adjusting total actual demand of power by the distributed computing facility by adjusting, via the processor, use of local energy generation or energy storage, or both, at the distributed computing facility.

21 . The method of claim 14 , comprising adjusting, via the processor, use of local energy generation or energy storage, or both, at the distributed computing facility in response to changing demand or changing pricing of energy, or a combination thereof.

22 . The method of claim 14 , comprising adjusting, via the processor, total actual demand of power in response to an incentive program or spot pricing of energy, or a combination thereof.

23 . A non-transitory, computer-readable medium comprising instructions executable by a processor to:
forecast power consumption by a distributed computing facility, comprising to:
forecast sequence of jobs execution on a system of the nodes over time;
estimate power for the jobs of the system; and
develop a system-level forecast of power of the system, wherein the system-level forecast is correlative with the sequence of job execution and the power for the jobs.

24 . The non-transitory, computer-readable medium of claim 23 , wherein to forecast power consumption comprises to forecast power consumption in response to a job event or an energy event, wherein the job event comprises submission of a job for execution to the system, completion of a job on the system, suspension of a job on the system, or any combination thereof.

25 . The non-transitory, computer-readable medium of claim 23 , comprising instructions executable by the processor to adjust use of local energy generation or energy storage, or both, at the distributed computing facility in response to changing demand of power at the distributed computing facility or changing pricing of energy, or both, and wherein to forecast power consumption comprises to develop a forecast of total actual demand for the distributed computing facility.

Assignees

• Intel Corp

Inventors

• Bodas Devadatta V
• Rajappa Muralidhar
• Song Justin J
• Hoffman Andy

Classifications

• G06F11/3006 Primary

  where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems (multiprogramming arrangements G06F9/46; allocation of resources G06F9/50) · CPC title

• G06Q10/06

  Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling · CPC title

• G06F11/3062

  where the monitored property is the power consumption (power management in a computing system G06F1/3203) · CPC title

• G06F2201/86

  Event-based monitoring · CPC title

• G01R21/00 Primary

  Arrangements for measuring electric power or power factor (G01R7/12 takes precedence) · CPC title