Discussion on row-level delete performance for CDC use case.

[Puneet Zaroo <pz...@netflix.com.INVALID>]

Hi all,
Our initial experience with using the row-level delete feature to support
streaming CDC, has shown lower than expected query and compaction
performance. I was wondering if anybody has got it working with good
performance for data-sources of lets say, at-least a few TB in size and
100s of mutations per second. If someone has got it working, I would love
to chat and learn more details of how the CDC pipelines / destination
iceberg tables were configured to get it working.
Thanks in advance,
- Puneet