Github user jerrypeng commented on the issue:
https://github.com/apache/storm/pull/2400
@revans2 thanks for the explanation! Sorry my math was wrong! I understand
the context a lot better now! Though my intuition tells me that if larger
topologies usually get scheduled first, this will likely to result in more
fragmentation and less topologies to be scheduled overall if there is a lot of
resource contention.---
