Hacker News Reader

I could be wrong here, I'm not an expert with databases, but I believe -- this is a belief, please help me disconfirm it if you think it is wrong -- that CitusDB's approach that shards Postgres across redundant mirrors, should successfully scale horizontally for OLTP without sacrificing consistency.

View on HN

Citus still has a centralized coordinator that all requests go through to be routed to their shard, or at least it did several years ago when I last ran it in production. It definitely scales further than single instances, but it's not a magic bullet.

Others can do it too: Spanner, FoundationDB. SI just has generally better performance because there is less to check, and you can always opt-in to row locking if you need (because of write-write conflicts)

Reply on news.ycombinator.com