On Availability of Intermediate Data in Cloud Computations

This paper takes a renewed look at the problem of managing intermediate data that is generated during dataflow computations (e.g., MapReduce, Pig, Dryad, etc.) within clouds. We discuss salient features of this intermediate data and outline requirements for a solution. Our experiments show that existing local writeremote read solutions, traditional distributed file systems (e.g., HDFS), and support from transport protocols (e.g., TCP-Nice) cannot guarantee both data availability and minimal interference, which are our key requirements. We present design ideas for a new intermediate data storage system.

• Type: Conference
• Year: 2009
• Where: HOTOS
• Authors: Steven Y. Ko, Imranul Hoque, Brian Cho, Indranil Gupta
• Keywords: Dataflow Computations, HOTOS 2009, Intermediate Data, Operating System, Traditional Distributed File