Research And Application Of Redis Memory Database In Power Trading

With the deepening reform of electric power industry in China,the marketization of electric power products is becoming more and more common,which is due to the steady construction of power trading center.As the key technology support of power trading center,power trading system needs to meet the diversified mode of power market and restore the commodity property of power.In power matched trading scenarios,the concurrency is very high and request for a low latency in centralized declaration period.Along with the increase in quantity of company in electricity market,power trading system faced a lot of test in data processing ability.Based on the requirements of developing power matched trading system,we solve the problem that concurrent throughput for handling data of the system is low and request delay is high in reporting period by using Redis memory database to provide data storage services.By studying the various features of Redis kernel,the Redis database is analyzed and optimized,and then it is applied to the data processing layer of power matched trading system to improve the system performance.At first,we consider the declared data needs to be sorted by time and price when they are stored,add the queue data types(dzqueue)in Redis kernel by using module function of Redis which can sort data by double attributes,ensure great system performance of processing data and reduce the difficulty of the development in data processing layer.And then studied master-slave replication strategy and analysis the problem of data consistency of Redis cluster,to meet master and slave node lost connection caused full synchronization problem,put forward the method of dynamic adjustment copy buffer based on the amount of access of the client,the method reduce the amount of the number of full synchronization.At the same time,strong data consistency scheme based on Quorum mechanism is proposed to solve final consistency problem of Redis command propagation.Research the Sentinel mechanism of Redis cluster and analyzes the key factors of improving the cluster availability,we put forward a method to select the data node based on the load which can ensure the probability of elected node failure is smaller as master and does not increase failure recovery time of system,improve the cluster's availability.Finally,the power matched trading system is realized for the project demand,which enables the power trading system to support the business scenarios of power matched trading.By testing and analyzing the optimized Redis database,the results of above optimization were effective.At the same time,compared with the traditional database,using Redis database to store data increases the speed of reading and writing requests of the system by an order of magnitude.