A Survey on Map-Reduce is a distributed Computational algorithm

Abstract

   Map/Reduce is a distributed computational   algorithm, which is originally designed by Google, Mapreduce is expanding in popularity and is being utilized for many large-scale jobs. The open-source Hadoop system has the most common implementation of Map/Reduce. For the fundamental storage backend, Hadoop by default manages the Distributed File System (HDFS), however Hadoop originally was planned to be compatible with other FS (file systems). Apart from HDFS  Hadoop does provide few other types of FS i.e. KFS, S3. Hadoop uses Java interface provided by these file systems. Lustre doesn’t contain JAVA wrapper. Lustre doesn’t accept like hadoop does. Lustre provides a POSIX-complaisant interface for UNIX file system. Common problems with Hadoop plus HDFS as a platform can be solved with Lustre as a backend system.