Search before asking
Motivation
When hadoop s3 filesystem is trying to upload a file larger than 128MB, it would convert from put object into multipart upload and sometimes it would stuck, especially during full compaction, Paimon will upload more files to object storage and being throttled.
Iceberg has provided an options write.object-storage.enabled to add a computed hash component for data path to prevent being throttled.
https://iceberg.apache.org/docs/nightly/docs/configuration/?h=write.object+storage.enabled#write-properties
It would be better if Paimon also provide the same fuctionality.
Solution
- adding computed hash for data files and leverage the ExternalPathProvider
Anything else?
No response
Are you willing to submit a PR?