Github user revans2 commented on the issue:
https://github.com/apache/storm/pull/2723
@danny0405 There is a lot of code here and I am trying to understand what
exactly this is doing, because it is hard to get it from just the code. I can
see the extra caching that you put in for the StormBase, which looks great.
The JIRA says that we are only scheduling topologies that need to be scheduled,
but I see that you had to change the isolation scheduler to disable that
feature. What exactly are the differences that the scheduler sees.
I am also a little confused about how this, besides the caching, improves
performance. All of the schedulers will loop through the list of topologies
and only schedule the ones that need to be scheduled. It is a tiny check that
takes a few ms at the most.
---
