Research on key technology of intelligent dispatching cloud data center based on Hadoop

Research on key technology of intelligent dispatching cloud data center based on Hadoop

Electric Yang Yanji

This paper focuses on three key problems of intelligent power grid dispatching cloud data center. First, the Cloud Data center internal network structure. Based on the analysis and comparison of traditional network structure algorithm and the newly proposed BCube and Dcell data center network structure algorithm, a more scalable and fault tolerant Pcube network structure algorithm is proposed. Second, the copy storage strategy of Hadoop cloud platform. Aiming at the shortcoming of the existing copy storage strategy of Hadoop, this paper studies the HDFS of improving the storage strategy--cohadoop copy storage strategy, and based on the FP array mining algorithm to define the correlation of locator. The performance of Cohadoop algorithm is perfected, which makes it better suited to the large and sparse transaction data set in intelligent power grid dispatching system. Third, the task scheduling strategy of Hadoop cloud platform. In this paper, the Fair scheduling algorithm of the existing Hadoop platform is improved, and the task selection is optimized based on the simulated annealing strategy, which is suitable for the task of intelligent power grid dispatching system and the distinction of priority. Finally, in order to verify the validity of the correlation algorithm, with the research process, this paper simulates the network structure algorithm, and verifies its extensibility and fault-tolerant characteristics. After that, to verify the performance of the Hadoop replica storage policy and the task scheduling algorithm, This paper builds Hadoop cluster with an existing cloud computing platform, simulates the related experimental environment and runs the relevant algorithm example, verifies the validity of the algorithm.

Research on key technology of intelligent dispatching cloud data center based on Hadoop