zhongyujiang commented on issue #3627:
URL: https://github.com/apache/paimon/issues/3627#issuecomment-2211612314

   Hi @BsoBird 
   
   Tired storage is one of the application scenarios. Implementing 
multi-location management has additional benefits, such as enabling a smoother 
transition of data from Hadoop to cloud object storage. Of course, this also 
requires support for multi-location metadata management.
   
   > You just need to regularly set the hdfs directory to a different storage 
policy, and then perform a compaction can be.
   
   Are you referring to Paimon data compaction? I think that’s something we 
want to avoid. Compaction requires a complete rewrite of the data, and the 
overhead of decoding and encoding is not trivial.
   
   
   > When is the best time to do your data migrated to this path, it becomes a 
problem.There also seems to be a consistency issue to consider, what if for 
some other reason the same data files appear in both paths?It's an unavoidable 
question.
   
   I believe data administrators can perform data migration operations based on 
demand. Data consistency is indeed a consideration, it can be achived by using 
atomic swaps. I don’t think this is an unavoidable issue.


-- 
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