This. > In this case I think you can let them become inconsistent in the face of, e.g., write errors. For devs using CF Durable Objects, people frequently use

This. > In this case I think you can let them become inconsistent in the face of, e.g., write errors. For devs using CF Durable Objects, people frequently use CF Queues or CF Workflows to ensure that everything is eventually consistent without significant overhead. It's a similar pattern to what large cos already do at scale with keeping data up to date across multiple partitions with Cassandra/DynamoDB.