Ruby Forum NGINX > $request_time granularity

Hello,

What part of the nginx workflow does $request_time cover?  I am guessing
that it is from the initial connect (or keep-alive) to the last write of 
the
response.  I am seeing a wide variance of times in my logs, and I am
guessing that it is due to high-latency clients versus nginx internals. 
Is
this a reasonable assumption?

Thanks.