Per a discussion on the Fly.io Community forum, it could be useful to introduce a replication delay in order to reduce bandwidth costs. If replicas only need updates on an infrequent basis (e.g. every minute, hour, or day), changes could be batched up and compacted into a single LTX file. That means multiple changes to a single page over the course of that time period could be batched up into a single page being replicated.
Per a discussion on the Fly.io Community forum, it could be useful to introduce a replication delay in order to reduce bandwidth costs. If replicas only need updates on an infrequent basis (e.g. every minute, hour, or day), changes could be batched up and compacted into a single LTX file. That means multiple changes to a single page over the course of that time period could be batched up into a single page being replicated.