Hacker News Reader

Well, as a question fairly on topic:

Has anyone successfully used CockroachDB in production?

As it seems to be the only viable alternative to spanner at the moment.

We use it as an insert only db for now for incoming webhooks which allows us to de-duplicate events and significantly reduce writes to the main PG db. We're running it on GKE which has been much easier to setup than PG.

There is also TiDB which is similar in some ways to CRDB: https://pingcap.com/en/

Have you had any major issues with it? It yes, how'd you dealt with them?

No major issues. If you're moving from a single node PG to CRDB you should expect CRDB to be slower because of all the consensus stuff. However almost every update I see from them contains some improvements in that area so it will get better with time probably. You just need to figure our if you're willing to accept a 2ms - 1m (PG) query execution vs 20ms flat execution (CRDB) - even after all the optimizations.

[cockroach here] You can check out stories of cockroachdb production deployments here -- https://www.cockroachlabs.com/customers/

Reply on news.ycombinator.com