mapreduce

Presented By:

Christopher Toste

Large Data is slow

MapReduce is a method of data processing for large data sets.

Think functionally!

Server CLUSTERS

Okay great... but what about MapReduce

Map(list(Key, Value)):
    for each K in Value:
        EmitIntermediate(list(K, "1"));  

Reduce(list(Key, Iterator)):
    int result = 0;
    for each K in Iterator:
        result += ParseInt(K);
    Emit(result);

This method is distributed

Questions?

Sources