Computer and data center load determination

• US 9,946,815 B1
• Filed: 12/02/2013
• Issued: 04/17/2018
• Est. Priority Date: 06/08/2007
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google LLC (Alphabet Inc.)
• Inventors: Weber, Wolf-Dietrich, Fan, Xiaobo, Barroso, Luiz Andre
• Primary Examiner(s): Myers, Paul R
• Application Number:
  US14/094,464
• Time in Patent Office: 1,597 Days
• Field of Search: 702 60, 702 57, 713200, 703 1
• CPC Class Codes: G06F 1/00 Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00 architectures of general purpose stored program computers G06F15/76
• CPC Class Codes: G06F 1/26 Power supply means, e.g. regulation thereof for memories G11C
• CPC Class Codes: G06F 1/3206 Monitoring of events, devices or parameters that trigger a change in power modality
• CPC Class Codes: G06F 2119/08 Thermal analysis or thermal optimisation
• CPC Class Codes: G06F 30/13 Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
• CPC Class Codes: G06F 30/20 Design optimisation, verification or simulation optimisation, verification or simulation of circuit designs G06F30/30
• CPC Class Codes: G06F 9/5094 where the allocation takes into account power or heat criteria power management in computers in general G06F1/3203; thermal management in computers in general G06F1/206
• CPC Class Codes: H05K 7/20836 Thermal management, e.g. server temperature control
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A method for use in deploying computers into a data center includes calculating in a computer an expected peak power draw for a plurality of computers. The expected peak power draw for the plurality of computers is less than a sum of individual expected peak power draws for each computer from the plurality of computers.

Claims

1. A method of estimating peak power usage of a computing system having a CPU, comprising:

identifying a computing system among a plurality of computing systems installed in a data center;

determining a plurality of components in the computing system;

determining a peak power usage value for each component of the plurality of components as the computing system is operating at an operating condition that comprises an application being performed by the computing system;

adding the peak power usage value for each component of the plurality of components together to result in an actual peak power value;

based on the resulting actual peak power value, aggregating actual peak power draws of the plurality of computing systems to determine an actual peak power draw of the data center that comprises the plurality of computing systems;

determining that the actual peak power draw of the data center is less than a sum of a maximum rated power draw of each of the plurality of computing systems;

based on the determination, deploying an additional plurality of the computing systems in the data center, wherein a sum of the maximum rated power draw of each of the plurality and additional plurality of computing systems is greater than a design power capacity of the data center;

determining a design power density of the data center based on the maximum rated power draw of each of the plurality and additional plurality of computing systems;

determining an average power density of the data center based on the aggregated actual peak power draws of each of the plurality and additional plurality of computing systems; and

determining an oversubscription ratio of the data center based on a ratio of the design power density to the average power density for the plurality and additional plurality of computing systems.

2. The method of claim 1, wherein the system comprises a server.

3. The method of claim 1, further comprising:

designing a power system for a rack of a portion of the plurality of the computing systems, the power system for the rack having a design power capacity greater than a sum of the peak power usage values of the portion of the plurality of the computing systems in the rack; and

designing a power system for the data center that comprises a plurality of the racks, the power system for the data center having the design power capacity less than a sum of the design power capacities of the plurality of the racks.

4. The method of claim 3, wherein the design power capacity of the data center is less than a sum of the actual peak power values of the plurality of computing systems in the data center.

5. The method of claim 1, wherein one of the components comprises a CPU, the method further comprising:

determining a peak CPU utilization;

estimating the peak power usage value of the CPU based on the peak CPU utilization.

6. The method of claim 1, further comprising determining a spatial layout of the data center based on the oversubscription ratio and the design power density.

7. The method of claim 1, further comprising determining a length of a row of one or more racks that support the plurality and additional plurality of computing systems in the data center based on the design power capacity, the design power density, and the oversubscription ratio.

8. The method of claim 7, wherein determining the length of the row comprises dividing the design power capacity by the design power density and multiplying by the oversubscription ratio.

9. A method of modeling computer system power usage for a single system, comprising:

generating a model of a power model of a running a group of computing systems, wherein each computing system has a CPU;

measuring total system power usage of each computing system in the group;

measuring CPU utilization of each computing system in the group;

determining a line fit of the system power to CPU utilization to create the model, wherein the line fit is based, at least in part, on the CPU utilization measured at an operating condition of the computing systems, a power usage of each computing system at the operating condition, and a power usage of each computing system at an idle state, the operating condition comprising an application performed by the computing systems;

based on the created model, determining a data center design power capacity for a plurality of groups of computing systems;

based on the created model, determining an oversubscription ratio of a data center based on a ratio of the design power density of the plurality of groups of computing systems to the average power density for the plurality of groups of computing systems;

based on the determined data center design power capacity and oversubscription ratio, determining a spatial layout of the data center, wherein a sum of the peak power draws of the plurality of groups of computing systems is greater than the determined data center design power capacity, and a sum of the average power draws of the plurality of groups of computing systems is less than the determined data center design power capacity; and

constructing the data center with the determined spatial layout.

10. The method of claim 9, wherein each group of computing systems comprises a rack of computing systems, each rack having a peak power supply capacity that is greater than a sum of the peak power draws of the computing systems in the rack.

11. The method of claim 9, wherein the application is a first application, and running a group of computing systems comprises:

running a first portion of the group of computing systems with the first applications; and

running a second portion of the group of computing system with a second application different than the first application.

12. The method of claim 9, wherein each computing system comprises a server.

13. A method of determining a number of computing systems that can be run in a data center, comprising:

determining a total power available; and

generating a model of computing system power usage for a single system, the generating comprising;

running a group of computing systems, wherein each computing system has a CPU;

measuring total system power usage of the computing system;

measuring an actual CPU utilization of each computing system in the group;

determining a line fit to the system power to the actual CPU utilization to generate the model, wherein the line fit is based, at least in part, on the CPU utilization measured at an operating condition of the computing systems, a power usage of each computing system at the operating condition, and a power usage of each computing system at an idle state, the operating condition comprising an application performed by the computing systems;

based on the created model, determining a data center design power capacity for a plurality of groups of computing systems; and

based on the created model, determining an oversubscription ratio of a data center based on a ratio of the design power density of the plurality of groups of computing systems to the average power density for the plurality of groups of computing systems;

based on the determined data center design power capacity and the oversubscription ratio, determining a spatial layout of the data center, wherein a sum of the peak power draws of the plurality of groups of computing systems is greater than the determined data center design power capacity, and a sum of the average power draws of the plurality of groups of computing systems is less than the determined data center design power capacity;

selecting a number of computing systems based on the generated model and an estimated amount of CPU utilization over a given period; and

constructing the data center with the determined spatial layout and the selected number of computing systems.

14. The method of claim 13, wherein selecting the number of computing systems based on the generated model and an estimated amount of CPU utilization over a given period comprises:

grouping computing systems into first tier groups and second tier groups, wherein a total system power usage of a computing system in the first tier group is different from a total system power usage in the second tier group; and

selecting a combination of first tier group computer systems and second tier group computer systems.

15. The method of claim 13, wherein each group of computing systems comprises a rack of computing systems, each rack having a peak power supply capacity that is greater than a sum of the peak power draws of the computing systems in the rack.

16. The method of claim 13, wherein the application is a first application, and running a group of computing systems comprises:

running a first portion of the group of computing systems with the first applications; and

running a second portion of the group of computing system with a second application different than the first application.

17. The method of claim 13, wherein each computing system comprises a server.