Optimizing energy use in a data center by workload scheduling and management

Patent metadata

Field	Value
Publication number	US-9235441-B2
Application number	US-201213709202-A
Country	US
Kind code	B2
Filing date	Dec 10, 2012
Priority date	Jan 10, 2011
Publication date	Jan 12, 2016
Grant date	Jan 12, 2016

Abstract

Techniques are described for scheduling received tasks in a data center in a manner that accounts for operating costs of the data center. Embodiments of the invention generally include comparing cost-saving methods of scheduling a task to the operating parameters of completing a task—e.g., a maximum amount of time allotted to complete a task. If the task can be scheduled to reduce operating costs (e.g., rescheduled to a time when power is cheaper) and still be performed within the operating parameters, then that cost-saving method is used to create a workload plan to implement the task. In another embodiment, several cost-saving methods are compared to determine the most profitable.

First claim

What is claimed is:

1. A method of managing tasks in a data center, comprising:
receiving a task to be performed by computing resources within the data center; determining, by operation of one or more computer processors, at least two cost-saving methods for scheduling the task based on a job completion constraint, wherein the job completion constraint defines one of (i) an operating parameter of the data center when executing the task and (ii) a penalty associated with executing the task;
estimating a required time to complete the task using the computing resources in the data center without the cost-saving methods;
comparing the time needed to complete the task to the job completion constraint; upon determining that the required time satisfies the job completion constraint, estimating a time needed to complete the task using the computing resources in the data center with the cost-saving methods;
identifying any penalty incurred by the cost-saving methods;
determining a profit of the cost-saving methods based on savings and any incurred penalty associated with the cost-saving methods;
determining the most profitable cost-saving method based on the profit of each cost-saving methods; and
upon determining the estimated time needed for the cost-saving methods satisfies the job completion constraint, scheduling the task using the most profitable of the cost-saving methods, wherein the most profitable cost-saving method delays the execution of the task in order to reduce the energy consumed by the data center relative to executing the task without a delay.

2. A method of managing tasks in a data center comprising:
receiving a task to be performed by computing resources within the data center;
determining, by operation of one or more computer processors, at least two cost-saving methods for scheduling the task based on a job completion constraint, wherein the job completion constraint defines one of (i) an operating parameter of the data center when executing the task and (ii) a penalty associated with executing the task, wherein the cost-saving methods generate savings by reducing the operating expenses of the data center when executing the task;
estimating a required time to complete the task using the computing resources in the data center without the cost-saving methods;
comparing the time needed to complete the task to the job completion constraint;
upon determining that the required time satisfies the job completion constraint, estimating a time needed to complete the task using the computing resources in the data center with the cost-saving methods;
identifying any penalty incurred by the cost-saving methods;
upon determining that the plurality of cost-saving methods do not incur a penalty, determining a profit of each cost-saving method based on the savings from each cost-saving method;
upon determining that at least one of the cost-saving methods does incur a penalty, determining the profit of the cost-saving methods based on the savings and any incurred penalty associated with the cost-saving methods;
determining the most profitable cost-saving method based on the profit of each cost-saving method; and
upon determining the estimated time needed for the cost-saving methods satisfies the job completion constraint, scheduling the task according to the most profitable cost-saving method, wherein the most profitable cost-saving method delays the execution of the task in order to reduce the energy consumed by the data center relative to executing the task without a delay.

3. The method of claim 1 , wherein the job completion constraint is set by at least one of: a contractual agreement and preferences of an administrator.

4. The method of claim 1 , wherein the data center is connected to a plurality of data centers within a cloud computing network.

5. The method of claim 4 , further comprising transmitting the task to a different data center within the cloud computing network.

6. The method of claim 1 , wherein the job completion constraint defines both (i) an operating parameter of the data center when executing the task and (ii) a penalty associated with executing the task.

7. The method of claim 1 , wherein estimating the time needed to complete the task with the cost-saving methods comprises estimating a number of data center units per a unit of time required in order to complete the task.

Assignees

• IBM

Inventors

• Brech Brad L
• Devarakonda Murthy V
• Lehman Bret W
• Ramos Stacey

Classifications

• Y02B60/144

  Cross-Sectional Technologies · mapped topic

• G06F9/4893 Primary

  taking into account power or heat criteria (power management in computers in general G06F1/3203; thermal management in computers in general G06F1/206) · CPC title

• Y02D10/00

  Energy efficient computing, e.g. low power processors, power management or thermal management · CPC title