From the analysis above we can see that the vertexes
in the same stage have the same distance to the
initial vertex which makes them have the same
priority in running and thus are splittable in
computing. As is shown in Figure 6, according to the
new method for splitting, we designed the procedure
for every vertex under Map Reduce. The pseudo code
of the procedure is also described in Table 1. The
BFS Mapper just re-emit the pair that it
receives. The nodes that have the same vertex ID
will be processed in the same Reducer, thus each
Reducer will receive a single node representing the
vertex and many nodes representing visiting
messages sent by its adjacent vertexes. If this vertex
has already been visited then these visiting messages
will be discarded, otherwise this vertex will be
visited during which the new visiting messages from
this vertex to its adjacent vertexes will be emitted.
Finally the vertex node will be re-emitted.