We've implemented streaming replication on all supported node client software. How does it work?

1. Master nodes handle negotiate with peers, sync the network and validate block data.
2. Instead of just writing validated block data to disk, it also writes it to a Kafka topic.
3. Replica nodes subscribe to that Kafka topic and write that data directly to disk, rather than syncing and peering to obtain new data.
4. Replicas—unlike Masters—respond to queries from Rivet subscribers.

This is what allows us to rapidly and easily add nodes to meet the capacity needs of an arbitrary number of subscribers operating at arbitrarily large or small volumes. And beyond that, streaming replication introduces a vast number of qualitative advantages to Rivet service as well. But don't take our word for it—try it out for yourself and see.