Dynamically placing computing jobs

• US 9,063,738 B2
• Filed: 11/22/2010
• Issued: 06/23/2015
• Est. Priority Date: 11/22/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Jain, Navendu, Menache, Ishai, Buchbinder, Niv
• Primary Examiner(s): TRUONG, LAN DAI T
• Application Number:
  US12/951,977
• Publication Number:
  US 20120130554A1
• Time in Patent Office: 1,674 Days
• Field of Search: 709/223, 709/224, 709/229, 709/226
• US Class Current: 1/1
• CPC Class Codes: G06F 1/329 by task scheduling
• CPC Class Codes: G06F 9/5027 the resource being a machine, e.g. CPUs, Servers, Terminals
• 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: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

This document describes techniques for dynamically placing computing jobs. These techniques enable reduced financial and/or energy costs to perform computing jobs at data centers.

Claims

1. A method implemented by one or more computer processing devices, the method comprising:

determining cost differences for electricity to perform a computing job at an individual data center having state information usable to perform the computing job and multiple remotely-located data centers, wherein the multiple remotely-located data centers are at different geographic locations than the individual data center and include at least;

a first remotely-located data center that does not have the state information, and

a second remotely-located data center that also does not have the state information;

determining multiple costs for bandwidth to communicate the state information from the individual data center to the multiple remotely-located data centers, wherein a first cost for bandwidth to communicate the state information to the first remotely-located data center of the multiple data centers is different than a second cost for bandwidth to communicate the state information to the second remotely-located data center;

determining further costs for bandwidth to communicate results of the computing job from the individual data center and the multiple remotely-located data centers to an entity that requested the computing job;

identifying, from the individual data center and the multiple remotely-located data centers, a selected data center to perform the computing job, wherein the selected data center is identified based on the cost differences for electricity, the multiple costs for bandwidth, and further costs for bandwidth; and

placing the computing job at the selected data center, wherein the determining the cost differences comprises calculating;

an individual expected electricity cost of performing the computing job using individual computational resources of the individual data center, and

multiple other expected electricity costs of performing the computing job using other computational resources of the multiple remotely-located data centers.

2. The method as recited in claim 1, further comprising:

prior to the determining the cost differences for electricity, receiving indications that the multiple remotely-located data centers are available to perform the computing job.

3. The method as recited in claim 1, wherein the determining the cost differences for electricity further comprises:

obtaining an indication of quantities of the individual computational resources that would be used to execute the computing job provided the computing job is placed at the individual data center;

obtaining other indications of other quantities of the other computational resources that would be used to execute the computing job provided the computing job is placed at the multiple remotely-located data centers, respectively;

obtaining an individual amount of electricity to perform the computing job using the quantities of the individual computational resources at the individual data center; and

obtaining other amounts of electricity to perform the computing job using the other quantities of the other computational resources at the multiple remotely-located data centers.

4. The method as recited in claim 1, wherein the determining the individual expected electricity cost is based on multiple energy sources available at the individual data center, at least one of the multiple energy sources being a renewable energy source.

5. The method as recited in claim 4, further comprising:

determining the individual expected electricity cost based on a projection of an availability of electricity from the renewable energy source at the individual data center when the computing job is performed.

6. The method as recited in claim 1, wherein the determining the cost differences for electricity is based on associated geographic locations of the individual data center and the multiple remotely-located data centers, respectively, the associated geographic locations affecting electricity costs at the individual data center and the multiple remotely-located data centers.

7. A system comprising:

one or more processing devices; and

one or more computer-readable memory devices storing program code that, when executed by the one or more processing devices, cause the one or more processing devices to;

determine a first electricity cost for electricity to perform a computing job using first computational resources of a first remotely-located data center, wherein the computing job has associated state information usable to perform the computing job,

determine a second electricity cost for electricity to perform the computing job using second computational resources of a second remotely-located data center,

determine a first bandwidth cost to communicate the associated state information to the first remotely-located data center and a second bandwidth cost to communicate the associated state information to the second remotely-located data center,

determine an individual data center from the first data center and the second data center for placement of the computing job, wherein the individual data center is determined based on the first electricity cost, the second electricity cost, the first bandwidth cost to communicate the associated state information to the first data center, and the second bandwidth cost to communicate the associated state information to the second remotely-located data center, and

place the computing job at the individual data center.

8. The system according to claim 7, wherein the individual data center is the lowest cost data center of the first data center and the second data center.

9. The system according to claim 7, wherein the computational resources of the first data center include multiple first servers and the computational resources of the second data center include multiple second servers.

10. The system according to claim 9, wherein the first data center and the second data center include different numbers of servers.

11. The system according to claim 7, wherein the first electricity cost is determined using a convex increasing function to represent a relationship between the first electricity cost and a number of the first computational resources used for the computing job.

12. One or more computer-readable memory devices storing program code which, when executed by one or more processing devices, causes the one or more processing devices to perform acts comprising:

determining a first electricity cost for electricity to perform a computing job locally using first computational resources of a first data center having state information for the computing job, wherein the state information is usable to perform the computing job;

determining a second electricity cost for electricity to perform the computing job using second computational resources of a second remotely-located data center that does not have the state information for the computing job;

determining a third electricity cost for electricity to perform the computing job using third computational resources of a third remotely-located data center that does not have the state information for the computing job;

determining a first bandwidth cost to communicate the state information from the first data center to the second remotely-located data center;

determining a second bandwidth cost to communicate the state information from the first data center to the third remotely-located data center, wherein the second bandwidth cost is different than the first bandwidth cost; and

identifying a selected data center from the first data center, the second remotely-located data center, and the third remotely-located data center for placement of the computing job, wherein the selecting considers;

the first electricity cost,

the second electricity cost,

the third electricity cost,

the first bandwidth cost, and

the second bandwidth cost.

13. The one or more computer-readable memory devices according to claim 12, the acts further comprising:

placing the computing job at the selected data center.

14. The one or more computer-readable memory devices according to claim 12, the acts further comprising:

in an instance when the second remotely-located data center is the selected data center, causing the first data center to send the state to the second remotely-located data center and not send the state to the third remotely-located data center.

15. The one or more computer-readable memory devices according to claim 12, wherein the first data center, the second remotely-located data center, and the third remotely-located data center each host different instances of an application configured to perform the computing job.

16. The one or more computer-readable memory devices according to claim 12, wherein the first electricity cost reflects both direct electricity costs and marginal electricity costs of operating the first computational resources.

17. The one or more computer-readable memory devices according to claim 12, wherein the determining the first electricity cost comprises modeling the first electricity cost based on a convex increasing function.

18. The one or more computer-readable memory devices according to claim 12, wherein the first electrical cost, the second electrical cost, and the third electrical cost each reflect different financial costs.

19. The one or more computer-readable memory devices according to claim 12, wherein the second electricity cost reflects an expected cost of a renewable energy resource at the second remotely-located data center based on projected weather conditions at the second remotely-located data center.

20. The one or more computer-readable memory devices of claim 12, wherein the first electricity cost and the second electricity cost are both determined using a function that accounts for both electricity sourced from at least one power grid and electricity sourced from at least one renewable energy source.