Techniques for orchestrated load shedding

Patent metadata

Field	Value
Publication number	US-12155210-B2
Application number	US-202318338962-A
Country	US
Kind code	B2
Filing date	Jun 21, 2023
Priority date	Nov 8, 2022
Publication date	Nov 26, 2024
Grant date	Nov 26, 2024

Abstract

Disclosed techniques relate to orchestrating power consumption reductions across a number of hosts. A number of response levels may be utilized, each having an association to a corresponding set of reduction actions. The impact to customers, hosts, and/or workloads can be computed at run time based on current and/or predicted conditions and workloads, and a particular response level can be selected based on the computed impact. These techniques enable a sufficient, but least impactful response to be employed.

First claim

What is claimed is:

1. A method, comprising:
determining, by a computer system, respective predicted sets of workloads executing on each of a plurality of hosts during a future time period;
identifying, by the computer system, a plurality of response levels that specify, based on one or more respective host attributes, applicability of a respective set of reduction actions, wherein a first response level of the plurality of response levels specifies, based on a first host attribute, applicability of a first set of reduction actions, wherein a second response level of the plurality of response levels specifies, based on a second host attribute, applicability of a second set of reduction actions, and wherein the first set of reduction actions is more severe than the second set of reduction actions;
determining 1) a first estimated power reduction that is expected for the first response level if the first response level is applied and 2) a second estimated power reduction that is expected for the second response level if the second response level is applied, the first estimated power reduction being determined based at least in part on the respective predicted sets of workloads and the first host attribute with which the applicability of the first set of reductions actions is specified, the second estimated power reduction being determined based at least in part on the predicted respective sets of workloads and the second host attribute with which the applicability of the second set of reduction actions is specified;
selecting the first response level over the second response level based at least on the first estimated power reduction expected if the first response level is applied and the second estimated power reduction expected if the second response level is applied, wherein the first estimated power reduction that is estimated to result from application of the first response level is less than the second estimated power reduction that is estimated to result from application of the second response level;
identifying one or more workloads of the respective predicted sets of workloads that (a) are currently executing on the plurality of hosts and (b) would be affected by application of the first set of reduction actions to the plurality of hosts according to the selected first response level;
prior to the future time period, preemptively migrating the one or more workloads from respective hosts to other respective hosts of the plurality of hosts; and
causing application of the first set of reduction actions to the plurality of hosts according to the first response level.

2. The method of claim 1 , wherein determining the first estimated power reduction that is expected for the first response level comprises:
determining, based on the first host attribute, that the first set of reduction actions are applicable to a first subset of hosts of the plurality of hosts;
identifying the respective predicted sets of workloads executing on each of the first subset of hosts; and
determining a respective power consumption of each of the respective predicted sets of workloads executing on each of the first subset of hosts.

3. The method of claim 2 , wherein determining the first estimated power reduction that is expected for the first response level further comprises:
determining a sum of the respective power consumption of the respective predicted sets of workloads executing on each of the first subset of hosts as the first estimated power reduction that is expected for the first response level if applied.

4. The method of claim 2 , wherein determining the first estimated power reduction that is expected for the first response level further comprises:
determining respective estimated power consumption of the predicted respective sets of workloads executing on each of the first subset of hosts after application of the first set of reduction actions to the first subset of hosts; and
determining the first estimated power reduction expected for the first response level, if applied, further based on (a) the respective power consumption of the respective predicted sets of workloads executing on each of the first subset of hosts and (b) the respective estimated power consumptions of the respective predicted sets of workloads executing on each of the first subset of hosts after application of the first set of reduction actions to the first subset of hosts.

5. The method of claim 1 , further comprising:
determining a difference between a current value for an aggregate power consumption of the plurality of hosts and a value for an aggregate power threshold of the plurality of hosts; and
determining that the first estimated power reduction expected for the first response level is greater than the difference;
wherein selecting the first response level over the second response level is based at least on determining that the first estimated power reduction for the first response level is greater than the difference.

6. The method of claim 1 , wherein the respective predicted sets of workloads executing on each of the plurality of hosts and the plurality of response levels that specify, based on one or more respective host attributes, applicability of the respective set of reduction actions to the plurality of hosts are identified based at least in part on at least one of:
identifying a degradation or failure of a temperature control system for a physical environment including the plurality of hosts; or
identifying a government reduction in power supply; or
identifying an increase in external temperature.

7. A method, comprising:
identifying, by a computer system, respective predicted sets of workloads executing on each of a plurality of hosts during a future time period;
identifying, by the computer system, a plurality of response levels that specify, using respective host attributes, applicability of a respective set of reduction actions to the plurality of hosts, wherein a first response level of the plurality of response levels specifies, using a first host attribute, applicability of a first set of reduction actions, and wherein the first set of reduction actions is more severe than a second set of reduction actions associated with a second response level of the plurality of response levels;
determining a first predicted impact for the first response level if the first set of reduction actions is applied, the first predicted impact being determined based at least on (a) the respective predicted sets of workloads executing on each of the plurality of hosts and (b) a first set of hosts to which the first set of reduction actions applies, each of the first set of hosts being individually associated with the first host attribute;
wherein the first predicted impact is determined based on at least one of a priority of a host, or a number of affected hosts, a number of affected workloads, priority levels of one or more affected workloads, a number of affected customers, or priority levels of the affected customers;
selecting the first response level from the plurality of response levels based at least on the first predicted impact for the first response level if the first set of reduction actions is applied, wherein the first predicted impact estimated to result from application of the first response level is less than a second predicted impact estimated to result from application of the second response level;
identifying one or more affected workloads of the respective predicted sets of workloads that (a) are currently executing on the plurality of hosts and (b) would be affected by application of the first set of reduction actions to the plurality of hosts according to the selected first response level;
prior to the future time period, preemptively migrating the one or more affected w

Assignees

• Oracle Int Corp

Inventors

• Zeighami Roy Mehdi
• Kochar Sumeet
• Herman Jonathan Luke
• Huang Mark Lee

Classifications

• H02J2105/54

  according to a non-electrical condition, e.g. temperature · CPC title

• H02J2105/52

  for limitation of the power consumption in the networks or in one section of the networks, e.g. load shedding or peak shaving · CPC title

• H02J2103/30

  Simulating, planning, modelling, reliability check or computer assisted design [CAD] of electric power networks · CPC title

• H02J3/17 Primary

  Demand-responsive operation of AC power transmission or distribution networks · CPC title

• G06N20/00

  Machine learning · CPC title