Design Of Optimized Log Replication Distributed Consensus Algorithm OLR-Raft Based On Raft

With the rapid development of the Internet in recent years,although the storage capacity of a single computer is increasing every year,the traditional single computer to deal with the exponential growth Mass storage is also in-creasingly stretched with the Moore's Law going to the limit.Distributed systems can solve storage problems,but many errors such as power problem,disk error and network partition happened every day in the large-scale distributed system.Replication is the only method to solve the problem that the system may stop its servivce when errors happen.And thus the distributed consensus is the most important issue in distributed system.Most existing distributed consensus protocols are difficult to understand and difficult to implement.Raft is an easy-to-understand and esay-to-implement distributed consensus protocol.But is does not achieve a higer performance than others.And its performance is expected to be improved in actual applications.OLR-Raft(Optimized Log Replication Raft)based on the Raft achieve a higher performace than Raft.It mainly optimizes the log replication of Raft.OLR-Raft implements the leader election,fault recovery,role change,log synchronization and log compression and other important functions of Raft.And OLR-Raft also designed backward log addressing optimization,buffered log and append log parallelly.OLR-Raft is easy to understand,east to implement and it achieve a higer performance than original Raft.A key-value storage system used OLR-Raft shows that it has higher write speed,higher read speed,a little more write delay,less read delay and less recovery time than the system used Raft.OLR-Raft does achieve a higher performance than original Raft.