Performance studies of different Distributed File Systems

Project Description:

♦ Explore different file systems § HDFS and HBASE http://hadoop.apache.org/ HDFS is the file system and HBASE is the big table § Swift: http://swift.openstack.org/ § Ceph: http://ceph.com/ § Lustre: http://www.lustre.org/ ♦ Identify a feature vector (what features should be considered if a user needs to select a file system to use) § Access latency, access throughput, directory service latency, etc. § Consistency model and solutions, availability solutions, etc. § Other special features ♦ Use IOzone as a benchmark to measure the performance features of the file systems § http://www.iozone.org/

Ø Project steps ♦ Investigate and report the major characteristics of different distributed file systems ♦ Identify and report the feature vector of distributed file systems ♦ Implementation § Set up 4 file systems on a cluster § Modify IOzone so that

Please click on the PDF and then open RAW to view the Results .

Scripts will be shared on Request .

Thanks -Raja Mannem