So we shutdown the cluster in order to upgrade the network cards to 10Gbit on all machines. It worked flawlessly after the cluster came back up again. The same transformation is now taking around 5 mins every time and the load on the cluster is mild.   Not sure if the mismatch in network speeds between the nodes or some sort of a lumbo state that the cluster was in was the problem. Maybe it was even a combination of those two. Now it works as expected. Thanks for the help @mbigelow ... View more