Study on Scheduling and Fault Tolerance Strategy of MapReduce
Xi Xiong · Microelectronics & Computer · 2007
MapReduce is a kind of parallel programming model which can be used to process and generate large data sets. The strategies of scheduling and fault tolerance play an important role in the execution of MapReduce. After analyzing the execution of MapReduce, we get the scheduling and fault tolerance model. Based on this model, using some common strategies in P2P model, we propose three new scheduling strategies, and we get the new fault tolerance mechanism for each scheduling strategy after making some modifications to the original fault tolerance method of MapReduce.