sgcisco commented on issue #11118:
URL: https://github.com/apache/hudi/issues/11118#issuecomment-2089073931

   @ad1happy2go record key looks as `record_keys=["timestamp", "A", "B", 
"C"],`. Where `timestamp` is monotonically increasing in ms, `A` a string with 
a range of some 500k values, `B` is similar to `A`, `C` is max hundred values.
   We use `upsert` which is a default operation but we don't expect any updates 
on the inserted values.
   We tried `insert` but observed latencies were worse.
   
   Increasing partitioning granularity from daily to hourly seems help to 
decrease latencies but not to solve the problem completely.
   ![Screenshot 2024-05-01 at 22 07 
16](https://github.com/apache/hudi/assets/168409126/7cd6bd72-2ecb-4826-99f6-567481b234bc)
   
   In this case partitioning size goes down from 100Gb to 4.7Gb.
   
   > Are you seeing the disk spill during this operation, you can try 
increasing the executor memory to avoid the same.
   
   No,  over 15h running job
   
   ![Screenshot 2024-05-01 at 22 19 
07](https://github.com/apache/hudi/assets/168409126/c860312d-ed03-427d-aaa8-ca9bedcb0ed5)
   
   


-- 
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]

Reply via email to