[GitHub] [beam] allenpradeep commented on pull request #12255: [BEAM-9910]Add mutation byte counters to SpannerIO.write

[GitBox <gi...@apache.org>]

allenpradeep commented on pull request #12255:
URL: https://github.com/apache/beam/pull/12255#issuecomment-670202696


   > @allenpradeep given that this seems to be a per-element change, have we done a performance analysis to confirm that there's won't be a significant performance degradation due to this ?
   > 
   > LGTM other than that.
   
   Hi Cham,
   While dealing with a batch of mutations the size of the batch is already calculated. We are just passing this size as a KV pair through the pipeline and so there would be no cost here. But if a batch fails, then we end doing a recalculation of individual mutation as we commit it separately which is the added cost here. This may not be significant compared to the cost of making a RPC call per mutation.
   However, I'm planning to work with Niel on this once he is back.
   In my initial tests, I did not find performance degradation when the batches did not fail.


----------------------------------------------------------------
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:
users@infra.apache.org