The Study Of Dynamic Replication Strategy Across Geographic Nodes In Regional Cloud Environments

With the rapid development of cloud computing the way people use the Internetare changed profoundly. There are various ways to constitute clouds. When the scaleof the cloud is continually expanding to a regional cloud coverage geographicallydiverse areas, how to ensure high reliability, low-latency is an important researchtopic in cloud computing system.One idea to solve this problem is using distributed file system and replication. Inlarge distributed systems, replication is a general technique to improve systemredundancy and improve the efficiency of data access. The static replication adoptedin existing cloud storage distributed file system treat all data equally, it cannot meetvaried user’s interest. Frequent cross-regional remote access to data not only causingpressure on the backbone network between regions, but also a serious impact on theresponse speed of cloud services.In this paper, aiming at the geographically distributed networks in regional cloudsystem, a dynamic replication strategy across geographic nodes in regional cloudenvironments is proposed. By monitoring the needs of users and data parameters, thisstrategy can be dynamically adjusted. The strategy introduces the concept of fileremote popularity degree based on temporal locality and the concept of averagespan of data requested. It make the popular data cross geographic nodes can bespread dynamically and quickly, make it possible to access data locality, therebyreducing the backbone network bandwidth consumption across geographic nodes andimprove the efficiency of data access.