Reducing power consumption in a server cluster

Patent metadata

Field	Value
Publication number	US-10101798-B2
Application number	US-201514729044-A
Country	US
Kind code	B2
Filing date	Jun 2, 2015
Priority date	Sep 15, 2008
Publication date	Oct 16, 2018
Grant date	Oct 16, 2018

Abstract

A method of reducing power consumption of a server cluster of host systems with virtual machines executing on the host systems is provided. The method includes recommending host system power-on when there is a host system whose utilization is above a target utilization, and recommending host system power-off when there is a host system whose utilization is below the target utilization. Recommending host system power-on includes calculating impact of powering on a standby host system with respect to reducing the number of highly-utilized host systems in the server cluster. Recommending host system power-off includes calculating impact of powering off a host system with respect to decreasing the number of less-utilized host systems in the server cluster.

First claim

What is claimed is:

1. A method comprising:
recommending host system power-on when there is a host system in a server cluster of host systems whose utilization is above a target utilization, and
recommending host system power-off when there is a host system in the server cluster whose utilization is below the target utilization,
wherein recommending host system power-on includes calculating of powering on a standby host system with respect to reducing the number of highly-utilized host systems in the server cluster, with the impact of powering on calculated by simulating moving some virtual machines from highly-utilized host systems to the standby host system being recommended to be powered-on, including iterating through standby host systems, for each respective standby host system, by invoking a software module, which supports virtual machine resource constraints, in a “what-if” mode assuming the respective standby host system was powered on and quantifying an impact of powering on the respective standby host system,

wherein recommending host system power-off includes calculating impact of powering off a host system with respect to decreasing the number of less-utilized host systems in the server cluster, with the impact of powering off calculated by simulating moving all virtual machines from the host system, which is being recommended to be powered-off, to less-utilized host systems, including iterating through powered-on host systems, for each respective powered-on host system, by invoking the software module in the “what-if” mode assuming the respective powered-on host system was powered-off and quantifying an impact of powering off the respective powered-on host system; wherein
calculating the impact of powering on includes computing for the server cluster a sum based on a distance above the target utilization for each host system in the server cluster above the target utilization and then comparing the sum for the server cluster without the host system powered-on with the sum calculated for the server cluster with the host system powered-on under a simulation.

2. The method of claim 1 , further comprising determining the utilization of each host system as a ratio of demand to capacity for that host system.

3. The method of claim 1 , wherein computing and comparing are repeated for each standby host system in the server cluster to determine whether that standby host system should be recommended to be powered-on.

4. The method of claim 1 , wherein calculating impact of powering off includes computing for the server cluster a sum based on distance below the target utilization for each host system in the server cluster under the target utilization then comparing the sum for the server cluster with the host system powered-off under a simulation with the sum calculated for the server cluster with the host system kept powered-on.

5. The method of claim 4 , wherein computing and comparing are repeated for each powered-on host system in the server cluster to determine whether that powered-on host system should be recommended to be powered-off.

6. The method of claim 1 , wherein recommending host system power-off includes calculating host power-off cost, wherein factors involved in calculating the host system power-off cost include one or more of a loss of the host system's resources during power-down, power consumed during a power-down period, a loss of the host system's resources during a subsequent power-on operation, power consumed during a power-up period, and costs of migrating virtual machines back onto the host system.

7. The method of claim 6 , further comprising performing a cost benefit analysis of powering-off the host system compared to the host system power-off cost.

8. A non-transitory computer-readable storage medium containing program instructions for a method, wherein execution of the program instructions by one or more processors of a computer system causes the one or more processors to perform steps of the method comprising:
recommending host system power-on when there is a host system in a server cluster of host systems whose utilization is above a target utilization, and
recommending host system power-off when there is a host system in the server cluster whose utilization is below the target utilization,
wherein recommending host system power-on includes calculating impact of powering on a standby host system with respect to reducing the number of highly-utilized host systems in the server cluster, with the impact of powering on calculated by simulating moving some virtual machines from highly-utilized host systems to the standby host system being recommended to be powered-on, including iterating through standby host systems, for each respective standby host system, by invoking a software module, which supports virtual machine resource constraints, in a “what-if” mode assuming the respective standby host system was powered on and quantifying an impact of powering on the respective standby host system, wherein recommending host system power-off includes calculating impact of powering off a host system with respect to decreasing the number of less-utilized host systems in the server cluster, with the impact of powering off calculated by simulating moving all virtual machines from the host system, which is being recommended to be powered-off, to less-utilized host systems, including iterating through powered-on host systems, for each respective powered-on host system, by invoking the software module in the “what-if” mode assuming the respective powered-on host system was powered-off and quantifying an impact of powering off the respective powered-on host system; wherein
calculating the impact of powering on includes computing for the server cluster a sum based on a distance above the target utilization for each host system in the server cluster above the target utilization and then comparing the sum for the server cluster without the host system powered-on with the sum calculated for the server cluster with the host system powered-on under a simulation.

9. The non-transitory computer-readable storage medium of claim 8 , wherein the steps further comprise determining the utilization of each host system as a ratio of demand to capacity for that host system.

10. The non-transitory computer-readable storage medium of claim 8 , wherein computing and comparing are repeated for each standby host system in the server cluster to determine whether that standby host system should be recommended to be powered-on.

11. The non-transitory computer-readable storage medium of claim 8 , wherein calculating impact of powering off includes computing for the server cluster a sum based on distance below the target utilization for each host system in the server cluster under the target utilization then comparing the sum for the server cluster with the host system powered-off under a simulation with the sum calculated for the server cluster with the host system kept powered-on.

12. The non-transitory computer-readable storage medium of claim 11 , wherein computing and comparing are repeated for each powered-on host system in the server cluster to determine whether that powered-on host system should be recommended to be powered-off.

13. The non-transitory computer-readable storage medium of claim 8 , wherein recommending host system power-off includes calculating host power-off cost, wherein factors involved in calculating the host system power-off cost include one or more of a loss of the host system's resources during power-down, power consumed during a power-down period, a loss of the host system's resources during a subsequent power-on operation, power consumed duri

Assignees

• Vmware Inc

Inventors

• Gupta Alok Kumar
• Ji Minwen
• Mann Timothy P
• Mobashir Tahir
• Rencuzogullari Umit
• Shanmuganathan Ganesha
• Wang Limin
• Holler Anne Marie

Classifications

• H04L43/0876

  Network utilisation, e.g. volume of load or congestion level · CPC title

• G06F1/3203

  Power management, i.e. event-based initiation of a power-saving mode · CPC title

• Y02D10/17

  Cross-Sectional Technologies · mapped topic

• G06F1/3287 Primary

  by switching off individual functional units in the computer system · CPC title

• H04L43/04

  Processing captured monitoring data, e.g. for logfile generation · CPC title