Cross-data center energy-saving load balancing method based on regional temperature differentiation

Abstract

The invention provides a cross-data center energy-saving load balancing method based on regional temperature differentiation. The method comprises the steps that a cooling energy consumption equationcaused by heat conduction is determined by fitting; an energy consumption equation of each IT device operating normally in a data center is determined; a total energy consumption equation of the datacenter is determined; a scheduling vector of a cloud service request is defined; a workload dispatched to the data center is determined; a tail delay constraint based on a service level protocol is determined; a delay probability of each cloud service request is calculated; and a temperature-aware cloud service scheduling architecture is constructed. The cross-data center energy-saving load balancing method based on the regional temperature differentiation has the advantages that the total cooling consumption of the data center is reduced under the condition of ensuring that the cloud servicesatisfies the tail delay requirement of the service level protocol, and the cloud service can still satisfy the tail delay constraint based on the service level protocol at the same time.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a cross-data center energy-saving load balancing method based on regional air temperature differentiation. Background technique [0002] The energy efficiency of data centers has been greatly improved due to the use of more efficient electronic components in the infrastructure or the use of more advanced energy consumption technologies. However, the number and scale of cloud services are also increasing, resulting in an increase in energy consumption of data centers. The latest research shows that data center energy consumption in the United States has increased significantly over the past few years. In 2014, it accounted for approximately 1.8% of U.S. electricity consumption, an increase of approximately 4% from 2010-2014. It is expected that data center energy consumption will continue to rise in the near future, and it can be predicted that it will increase by 4% in ...

AI Technical Summary

Problems solved by technology

However, some other factors need to be considered when rescheduling these workloads, such as which data center the workload should be scheduled to be optimal; the rescheduled cloud service must still be able to meet the tail delay requirements; in order to provide high availability, The basic data required by cloud services is usually backed up between data centers, and backup is a basic feature provided by the data center infrastructure, so the energy consumption required for data backup is not considered

Images