| As a distributed real-time computing system,big data stream processing is simple and efficient to operate,and had been used in all walks of life.Meanwhile,as an open source,distributed real-time,distributed real-time computing system,Apache Storm has the high energy consumption.In order to solve this problem that Storm has the high energy consumption and low energy efficiency.This article is based on the basic framework and processing mode of Storm platform,defining a series of concepts and proposing some models of topological logic diagram,critical path,the data processing cost of critical path,ratio of performance-energy consumption,threshold of the volume of transmitted data and CPU utilization.An energy-saving regulation strategy for work nodes on Storm platform is proposed,which includes two energy-efficient algorithms aiming at whether there are any work nodes executing on the critical path of a topology.For the work nodes that executing on the critical path of a topology,EACP(Energy-efficient Strategy by adjusting the DRAM voltage of work nodes in critical path)is proposed.Under the constraints of the performance-consumption ratio model,the volume of transmitted data and CPU utilization,energy saving is achieved by adjusting the DRAM voltage of work nodes.For the work nodes that executing on the no-critical path of a topology,EANP(Energy-efficient Strategy by adjusting the DRAM voltage of work nodes in no-critical path)is proposed,without affecting the performance of system,the CPU voltage of work nodes is adjusted according to the volume of transmitted data and CPU utilization.The experimental results show that compared with the original system,the system which implements the energy-saving regulation strategy of the critical path can save the energy consumption by about 34.9%,and the system which implements the energy-saving regulation strategy of the no-critical path can save the energy consumption by about 49.5%. |
