June 24, 2014

Features:esProc parallel system VS MapReduce

MapReduce is a computing framework for job scheduling and fault tolerance in cluster computing; MapReduce is not involved in computing and does not offer basic function library.

esProc concerns with computing with rich function library; esProc does not offer framework, scheduling and fault tolerance can be done by developers via function library.

MapReduce is a complex framework particularly for large clusters, and the data exchange and fault tolerance method is inherent, which is difficult to tune in accordance with task features for better performance.

esProc can flexibly control job scheduling and balance performance and fault tolerance as per the task features, thus light-weight parallel computing can be realized in small and medium for higher performance.

No comments:

Post a Comment