clintropolis commented on issue #12262:
URL: https://github.com/apache/druid/issues/12262#issuecomment-1183707311

   >We don't have to keep the current request/response structure. If it helps, 
we can change to a protocol that allows multiplexing of multiple requests over 
the same connection.
   
   Sure, though that is probably an involved migration for current applications 
that use druid. The simplicity of the current API is attractive for interactive 
speed queries at least, so I'd be sad to see it leave completely.
   
   >Queues can't grow forever. Async handling allows the queue to grow bigger 
than request-per-thread handling, but there is still a limit, and we still need 
to have some reasonable behavior around those limits.
   
   Totally, the jetty request queue under the covers of what we are currently 
using is limited too, we just aren't in direct control of it or its contents 
and happens before we've parsed request bodies into queries so that we can 
understand and categorize them. That is why the only place where we can enforce 
our limits right now is on the active thread pool size which is dramatically 
smaller than the queue size. If we control the queue (queues), then we could 
limit based on queue size instead of the approximate active running query 
count, which i think is what we want.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to