Re: [PR] KAFKA-19389: Fix memory consumption for completed share fetch requests [kafka]

[via GitHub]

apoorvmittal10 commented on PR #19928:
URL: https://github.com/apache/kafka/pull/19928#issuecomment-2957325074

   @junrao The importance of the fix is more evident when messages are larger 
in size. I produced messages of size 102400 Bytes hence a single share fetch 
should have 500 (default) * 102400 = 50 MB max data. Ran with 5 share groups, 
with timeout of 5 minutes.
   
   The consumption bytes and records are similar, though with fix there is 
better perfromance i.e. ~8.7 GB read per group vs ~8.9 GB read with fix in 5 
minutes. Also the memory footprints are far lower with the fix as memory is 
reclaimed faster now.
   
   ### Without fix - trunk
   
   ![Screenshot 2025-06-10 at 00 38 
38](https://github.com/user-attachments/assets/a3e0696d-800a-4d5d-aa5e-2e3e22b95371)
   
   ![Screenshot 2025-06-10 at 00 38 
46](https://github.com/user-attachments/assets/ed2c9cfd-1e62-4410-8051-373efa83fa0e)
   
   ### With Fix
   
   ![Screenshot 2025-06-10 at 00 49 
27](https://github.com/user-attachments/assets/28410bdd-7bc7-4d02-abba-1637019f87ce)
   ![Screenshot 2025-06-10 at 00 52 
09](https://github.com/user-attachments/assets/9903b123-4469-440b-915c-63668667f99d)
   
   


-- 
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: jira-unsubscr...@kafka.apache.org

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