Cluster power consumption control method and device

Abstract

The invention provides a cluster power consumption control method and device. The power consumption control method comprises following steps: collecting the working parameter of each element of the servers in a cluster; building a power consumption model according to the working parameters; calculating the power consumptions of the servers; and making a power consumption strategy according to the power consumptions and executing the power consumption strategy. According to the method and the device of the invention, the power consumption model is built through collecting the working parameters of the servers and rationally analyzing the parameters; the actual power consumptions of the servers are calculated according to the power consumption model; further the rational strategy is made according to the actual power consumptions of the servers; the power consumptions of the servers and the cluster are adjusted through executing the strategy; the integrated power consumption of the server cluster is distributed rationally; and therefore, the efficiency is improved.

Description

technical field [0001] The present invention relates to the field of servers, in particular to a cluster power consumption control method and device. Background technique [0002] With the increasingly prominent energy problems worldwide, how to build energy-saving, environmentally friendly, highly reliable and available green data centers has become a major challenge for the computer industry. According to statistics, by 2012, global data centers spent up to 26 billion US dollars on electricity costs each year, equivalent to 1.5% of the world's energy, and this figure will double in 2014. Studies have shown that on average only 6% to 12% of all electricity consumed by data centers is used to power servers to perform computing tasks, while the remaining electricity is mainly consumed by servers that are idle and respond to new tasks that may arise at any time. [0003] In general, the applications running on each cluster in the data center are different and generally do not...

AI Technical Summary

Problems solved by technology

Power consumption management at the local node level uses dynamic voltage and frequency adjustment DVFS technology to perform CPU frequency modulation to reduce power consumption, but it does not take into account the power consumption of other components; another prior art proposes a cluster energy-saving system based on resource scheduling, Monitor the idle state of the server to shut down or sleep the idle server. The energy saving method of this system is relatively simple, and the real-time control of each component of the server is not considered.

Images