gcs-connector-for-apache-kafka
gcs-connector-for-apache-kafka copied to clipboard
[File size flush inteverval] Best practices for reducing network costs
Hi all, hope everithing is fine. Currently we are working on a feature that uses this connector to dump debezium CDC messages. Our deployed connector is currently working based on default values, and as far as I can get, the only ways to control flush intervals are based on time (offset.flush.interval.ms, which defaults to 60 secs) and num of records per flushed files (file.max.records, didn't see a default value, so I'm assuming it will write as many lines as possible to the file during the flush interval window).
Our current cost for writing these files is pretty large, so I was wondering if you'd have any bet practices to improve performance of the connectors in order to reduce network costs during the write to gcs.
Thanks for any help! Cheers