Accelerating big data analytics on HPC clusters using two-level storage.

•Develop a new two-level storage system that integrates an upper-level in-memory file system with a lower-level parallel file system.•Model and compare I/O throughput of two-level storage to HDFS and OrangeFS.•Build a prototype of two-level storage system with Tachyon and OrangeFS,•Conduct experiments on real systems show that the proposed two-level storage delivers higher aggregate I/O throughputs than HDFS and OrangeFS and achieves weak scalability on both read and write.