A low-energy scheduling system and method for web clusters

Abstract

The invention discloses a web-cluster-oriented low energy consumption scheduling system and a method. The scheduling system comprises an energy-sensible scheduling module, a hysteresis controller and a manager, wherein the energy-sensible scheduling module, the hysteresis controller and the manager are connected sequentially. By the aid of the scheduling system, the resource utilization ratio in a web cluster system is greatly improved, and the energy consumption of the system is reduced. According to the method, a load connection number in the web cluster system serves as a sensitive factor, the system can schedule cluster nodes and virtual machine resources through monitoring of the load connection number only, the method has the advantages of being simple to operate, easy to implement, capable of achieving real-time scheduling and the like, and the scheduling method can improve the sensitivity of resource scheduling of the system.

Description

technical field [0001] The invention relates to resource scheduling technology in a computer cluster system, in particular to a web cluster-oriented low-energy scheduling system and method. Background technique [0002] The study found that the main expenses of the computer data center are the machine power supply and cooling system. Due to the influence of sudden loads such as regional imbalance and time imbalance in the access to resources, the resource utilization rate of the entire data center is low. The actual change of the load is used to provide the most reasonable service, resulting in a waste of resources and increasing unnecessary economic expenses. In the data center, it consumes unnecessary power and increases the burden on cooling equipment. The data shows that in 2003, a data center in the United States used 22TWh of electricity for energy supply, network switching equipment, cooling, etc., while the electricity price at that time was 100 US dollars per megaw...

AI Technical Summary

Problems solved by technology

However, there are certain problems in monitoring physical sources. For example, when the CPU utilization rate changes weakly within a certain period of time, it has poor autoregressiveness and aperiodicity. In this case, the automatic migration strategy cannot bring In addition, the CPU is only one of the three major computer physical resources. It is quite difficult and the accuracy is not high to fully consider the three resources; moreover, they did not consider the capabilities of the virtual cluster system. consumption problem

Images