Does anyone have experience use MapReduce to solve network flow problem

MapReduce has been shown to be powerful solve problem with large data sets in a parallel/distributed way.

Some combinational optimization problem such as maximum network flow, mini开发者_如何学编程mum cost network flow, multi commodity minimum cost flows, or shortest distance path/path-pair problems are known to be able to scale to very large size.

Does anyone has successful/failure experience to apply MapReduce to handle these types of problem? Could you please share your opinion whether it is a good fit or bad idea to resolve to MapReduce to solve such type of problems?

---

Felix Halim and others published a paper in 2011 that discusses how they solve max flow problem using map reduce. They "are able to compute max-flow on graph with 411 million vertices and 31 billion edges using 21 machines in reasonable time"!