Virtual machine scheduling using optimum power-consumption profile
View Patent ↗Methods and systems of determining an optimum power-consumption profile for virtual machines running in a data center are disclosed. In one aspect, a power-consumption profile of a virtual machine and a unit-rate profile of electrical power cost over a period are received. The methods determine an optimum power-consumption profile based on the power-consumption profile and the unit-rate profile. The optimum power-consumption profile may be used reschedule the virtual machine over the period.
1. A system to schedule virtual machines comprising:
one or more processors;
one or more data-storage devices; and
a routine stored in the data-storage devices and executed using the one or more processors, the routine
determining an optimum power-consumption profile based on a power-consumption profile of a virtual machine over a period of time and a unit-rate profile of electrical power cost over the period of time, wherein the optimum power-consumption profile minimizes power consumption of the virtual machine by substantially aligning minima of the optimum power-consumption profile with maximum unit-rate cost of the unit-rate profile and substantially aligning maxima of the optimum power-consumption profile with minimum unit-rate cost of the unit-rate profile; and
rescheduling the virtual machine over the same period with a workload having the optimum power-consumption profile.
2. The system of claim 1 wherein the power-consumption profile represents power consumption of the virtual machine periodically scheduled over time.
3. The system of claim 1 , wherein workload associated with the optimum power-consumption profile equals workload associated with the power-consumption profile.
4. The system of claim 1 wherein the optimum power-consumption profile minimizes power cost of the virtual machine execution.
5. A system to schedule virtual machines comprising:
one or more processors;
one or more data-storage devices; and
a routine stored in the data-storage devices and executed using the one or more processors, the routine
determining an optimum power-consumption profile based on a power-consumption profile of a virtual machine over a period of time and a unit-rate profile of electrical power cost over the period, wherein determining the optimum power-consumption profile further comprises;
computing a total workload based on the power-consumption profile;
ordering the unit-rate profile to generate an order unit-rate profile;
forming an ordered unit-rate index set from indices of the ordered unit-rate profile;
initializing a minimum power-consumption profile based on a minimum power level of the power-consumption profile;
ordering power levels of the power-consumption profile from smallest to largest in an ordered power-consumption profile; and
iteratively updating power levels of the minimum power-consumption profile to obtain the optimum power-consumption profile based on the ordered unit-rate index set and the ordered power-consumption profile; and
rescheduling the virtual machine over the period with a workload having the optimum power-consumption profile.
6. The system of claim 5 wherein iteratively updating power levels of the minimum power-consumption profile to obtain the optimum power-consumption profile further comprises:
for each iteration,
computing a workload based on the minimum power-consumption profile;
computing an error difference between the total workload and the workload;
determining a maximum power level in power-consumption profile;
reading a power level from the ordered power-consumption profile based on the ordered unit-rate index set;
computing an increment based on a minimum of the error and a difference between the maximum power level and the power level from the ordered power-consumption profile;
summing the power level from the ordered power-consumption profile and the increment to obtain an updated power level for the minimum power-consumption profile; and
subtracting the maximum power level from the power-consumption profile.
7. A method of reducing power consumption of data center comprising:
determining an optimum power-consumption profile based on a power-consumption profile of a virtual machine over a period of time and a unit-rate profile of electrical power cost over the period of time, wherein the optimum power-consumption profile minimizes power consumption of the virtual machine by substantially aligning minima of the optimum power-consumption profile with maximum unit-rate cost of the unit-rate profile and substantially aligning maxima of the optimum power-consumption profile with minimum unit-rate cost of the unit-rate profile; and
rescheduling the virtual machine over the same period with a workload having the optimum power-consumption profile.
8. The method of claim 7 wherein the power-consumption profile represents power consumption of the virtual machine periodically scheduled over time.
9. The method of claim 7 wherein workload associated with the optimum power-consumption profile equals workload associated with the power-consumption profile.
10. The method of claim 7 wherein the optimum power-consumption profile minimizes power cost of the virtual machine execution.
11. A method of reducing power consumption of data center comprising:
determining an optimum power-consumption profile based on a power-consumption profile of a virtual machine over a period of time and a unit-rate profile of electrical power cost over the period, wherein determining the optimum power-consumption profile further comprises;
computing a total workload based on the power-consumption profile;
ordering the unit-rate profile to generate an order unit-rate profile;
forming an ordered unit-rate index set from indices of the ordered unit-rate profile;
initializing a minimum power-consumption profile based on a minimum power level of the power-consumption profile;
ordering power levels of the power-consumption profile from smallest to largest in an ordered power-consumption profile; and
iteratively updating power levels of the minimum power-consumption profile to obtain the optimum power-consumption profile based on the ordered unit-rate index set and the ordered power-consumption profile; and
rescheduling the virtual machine over the period with a workload having the optimum power-consumption profile.
12. The method of claim 11 wherein iteratively updating power levels of the minimum power-consumption profile to obtain the optimum power-consumption profile further comprises:
for each iteration,
computing a workload based on the minimum power-consumption profile;
computing an error difference between the total workload and the workload;
determining a maximum power level in power-consumption profile;
reading a power level from the ordered power-consumption profile based on the ordered unit-rate index set;
computing an increment based on a minimum of the error and a difference between the maximum power level and the power level from the ordered power-consumption profile;
summing the power level from the ordered power-consumption profile and the increment to obtain an updated power level for the minimum power-consumption profile; and
subtracting the maximum power level from the power-consumption profile.
13. A non-transitory computer-readable medium encoded with machine-readable instructions that implement a method carried out by one or more processors of a computer system to perform the operations of:
determining an optimum power-consumption profile based on a power-consumption profile of a virtual machine over a period of time and a unit-rate profile of electrical power cost over the period of time, wherein the optimum power-consumption profile minimizes power consumption of the virtual machine by substantially aligning minima of the optimum power-consumption profile with maximum unit-rate cost of the unit-rate profile and substantially aligning maxima of the optimum power-consumption profile with minimum unit-rate cost of the unit-rate profile; and
rescheduling the virtual machine over the same period with a workload having the optimum power-consumption profile.
14. The medium of claim 13 wherein the power-consumption profile represents power consumption of the virtual machine periodically scheduled over time.
15. The medium of claim 13 wherein workload associated with the optimum power-consumption profile equals workload associated with the power-consumption profile.
16. The medium of claim 13 wherein the optimum power-consumption profile minimizes power cost of the virtual machine execution.
17. A non-transitory computer-readable medium encoded with machine-readable instructions that implement a method carried out by one or more processors of a computer system to perform the operations of:
determining an optimum power-consumption profile based on a power-consumption profile of a virtual machine over a period of time and a unit-rate profile of electrical power cost over the period, wherein determining the optimum power-consumption profile further comprises;
computing a total workload based on the power-consumption profile;
ordering the unit-rate profile to generate an order unit-rate profile;
forming an ordered unit-rate index set from indices of the ordered unit-rate profile;
initializing a minimum power-consumption profile based on a minimum power level of the power-consumption profile;
ordering power levels of the power-consumption profile from smallest to largest in an ordered power-consumption profile; and
iteratively updating power levels of the minimum power-consumption profile to obtain the optimum power-consumption profile based on the ordered unit-rate index set and the ordered power-consumption profile; and
rescheduling the virtual machine over the period with a workload having the optimum power-consumption profile.
18. The medium of claim 17 wherein iteratively updating power levels of the minimum power-consumption profile to obtain the optimum power-consumption profile further comprises:
for each iteration,
computing a workload based on the minimum power-consumption profile;
computing an error difference between the total workload and the workload;
determining a maximum power level in power-consumption profile;
reading a power level from the ordered power-consumption profile based on the ordered unit-rate index set;
computing an increment based on a minimum of the error and a difference between the maximum power level and the power level from the ordered power-consumption profile;
summing the power level from the ordered power-consumption profile and the increment to obtain an updated power level for the minimum power-consumption profile; and
subtracting the maximum power level from the power-consumption profile.