[chehai]

In order to get good query performance from Iceberg, we have to run compaction frequently. Compaction turns out to be very expensive. Any tip to minimize compaction while keeping queries fast?