zuston opened a new pull request, #2507:
URL: https://github.com/apache/uniffle/pull/2507

   
   ### What changes were proposed in this pull request?
   
   Introduce option to enable reorder multi servers for reader
   
   ### Why are the changes needed?
   
   If partition splitting is enabled, large partitions will be distributed 
across multiple shuffle servers. With the help of Spark AQE (Adaptive Query 
Execution), these large partitions will be processed by multiple tasks.
   
   In this case, all split tasks may sequentially read from the same set of 
shuffle servers, which can lead to high RPC concurrency pressure on specific 
servers.
   
   This PR introduces the ability to randomly reorder the underlying shuffle 
servers to achieve better load balancing during reading.
   
   ### Does this PR introduce _any_ user-facing change?
   
   Yes.
   
   `rss.client.read.reorderMultiServersEnable=false
   `
   ### How was this patch tested?
   
   Needn't
   


-- 
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