Replication delay configuration
First, thank you for providing read replicas. One thing that is still nessesary is the ability to tweak the replication configuration. If we want to take dumps from a medium (or larger) sized database directly from the replica, then this dump will be aborted because it simply takes too long.
This could be solved by enabling the max_standby_streaming_delay property for the read replicas. Setting it to -1 would enable to take dumps for larger databases of 100GB+.
Thank you for your feedback. max_standby_streaming_delay is now an available parameter in Azure Database for PostgreSQL. We also released hot_standby_feedback, max_standby_archive_delay, vacuum_defer_cleanup_age.
Rachel on behalf of the Azure Database for PostgreSQL team