Data and Parity block Placement Policy to enhance storage efficiency and utilization

摘要

HDFS which is known as Hadoop Storage manages fault-tolerance by data block replication. Replicas of data blocks are transmitted to other datanodes in a rack awareness manner. However prior HDFS storage management method wastes storage spaces and makes the situation of unbalanced cluster state. Proposed solution in this paper can save the storage space and maintain the balanced state of the entire cluster.