saintstack commented on pull request #1593:
URL: https://github.com/apache/hbase/pull/1593#issuecomment-623187675


   After reading the above:
   
   It was a mistake suggesting you generalize hedged reads Bharath -- at least 
for branch-2. I apologize if it made you do more work.
   
   On read-replicas in RPC tier, it strikes me that this an base-special in 
need of support from tiers above RPC? It won't fit?
   
   On the notion that the hedged reads addition to RPC implies our building a 
generic, feature-ful, RPC lib, let us take pause here (sorry if my suggestion 
that hedged reads be made generally available in RPC helped imply this). Lets 
not do this. Lets see if we can use an existing RPC lib instead (it may not be 
possible given the reminder above that retries, read-replicas, and hedging 
factor info from higher tiers). The Duo async refactor helped undo our RPC 
tangle.  If we could plug in a GRPC, perf allowing, it would be fun being able 
to just enable ‘pushback’, controlled delay, tracing, etc., because we are up 
on an RPC lib that just supports these abilities and others (May not be 
possible in the hbase context but worth investigation).
   
   Duo's reminder that we have to be careful when it comes to almost-there 
optional code is timely because too often, even with the best of intentions, 
the stuff is not enabled and rots.
   
   I'm game to help out here how ever I can.


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


Reply via email to