I'd like to keep track of the size of documents (just the actual fields,
not the whole JSON) that are getting indexed.
Parsing JSON on every request won't scale..
What would be the most elegant and yet fast way of doing that?
Some small % of error is OK.
This number doesn't have to be super up to date, but it shouldn't be more
than 5-10 minutes delayed either.
ELASTICSEARCH Performance Monitoring - http://sematext.com/spm/index.html