Big Data Simplified

Introducing MapReduce | 77

4.3 PARALLELISM IN MAP AND REDUCE PHASES

Let us now briey discuss the map and reduce processes which run on a cluster. It should be

quite apparent that, the more parallel processes we run, the higher is the degree of optimization.

We have seen that map and reduce operations operate on key value pairs. A raw dataset of

profile views in a social networking site, like LinkedIn, might include a record for every time a

member views another member’s profile. However, assume that the final data that we are looking

for from that dataset is the cumulative number of views of every member’s profile as shown in

Figure 4.8.

FIGURE ...