System and method for limiting the impact of stragglers in large-scale parallel data processing

A system for large-scale processing of data, comprising:

• memory;
• one or more processors;
• and
• one or more modules stored in the memory and executed by the one or more processors, the one or more modules including instructions to:
• execute a plurality of processes on a plurality of interconnected processors, the plurality of processes including a master process for coordinating a data processing job for processing a set of input data, and plurality of map processes and a plurality of reduce processes;
• in the master process, assign input data blocks of a set of input data to respective map processes of the plurality of map processes and assigning partitions of intermediate data to respective reduce processes of the plurality of reduce processes;
• in each of the plurality of map processes:
• execute an application-independent map program to retrieve a sequence of input data blocks assigned thereto by the master process and to apply an application-specific map function to each input data block in the sequence to produce the intermediate data;
• and
• store the intermediate data in memory of the interconnected processors;
• and
• in each of the plurality of reduce processes:
• receive a respective partition of the intermediate data from the memory of the interconnected processors;
• and
• apply an application-specific reduce function to the respective partition of the intermediate data to produce output values;
• and
• in a respective reduce process:
• receive multiple distinct partitions of the intermediate data and
• process the multiple partitions one at a time in succession;
• and
• identify the respective reduce process as a reduce process that is delaying the data processing job while continuing to process intermediate data and, in response, reassign at least one of the multiple partitions, which has not yet been processed, to a second reduce process, including copying the intermediate data in the reassigned partition to the other reduce process.

A non-transitory computer readable storage medium storing one or more programs for execution by one or more processors of a client device, the one or more programs comprising instructions to:

• execute a plurality of processes on a plurality of interconnected processors, the plurality of processes including a master process for coordinating a data processing job for processing a set of input data, and plurality of map processes and a plurality of reduce processes;
• in the master process, assign input data blocks of a set of input data to respective map processes of the plurality of map processes and assigning partitions of intermediate data to respective reduce processes of the plurality of reduce processes;
• in each of the plurality of map processes:
• execute an application-independent map program to retrieve a sequence of input data blocks assigned thereto by the master process and to apply an application-specific map function to each input data block in the sequence to produce the intermediate data;
• and
• store the intermediate data in memory of the interconnected processors;
• and
• in each of the plurality of reduce processes:
• receive a respective partition of the intermediate data from the memory of the interconnected processors;
• and
• apply an application-specific reduce function to the respective partition of the intermediate data to produce output values;
• and
• in a respective reduce process:
• receive multiple distinct partitions of the intermediate data and
• process the multiple partitions one at a time in succession;
• and
• identify the respective reduce process as a reduce process that is delaying the data processing job while continuing to process intermediate data and, in response, reassign at least one of the multiple partitions, which has not yet been processed, to a second reduce process, including copying the intermediate data in the reassigned partition to the other reduce process.