Migrate from single DC setup to multi DC geo replication in YugabyteDB

Viewed 3

[Question posted by a user on YugabyteDB Community Slack]

Right now, we want to start with a single data center YugabyteDB deployment. In the future, we plan to support multiple physical data centers that are geo replicated.

But we are not sure how-to seamlessly move forward from a single DC setup to an multi DC setup with geo replication. Is it possible to simply add an existing cluster to a geo-replicated setup and yugabyte automatically spread the data into the new DCs? Or what would be the best way to do so?

We are mainly using YCQL.

1 Answers

Short Answer, Yes.

You can add new nodes from other DC into the cluster and data will be replicated automatically and all nodes (from both DC) will allow read and write. So long as all nodes can communicate with each other at the network level, it should just work. Data replication will be synchronous in this case. So query time will include the network latency.

If the elevated transaction time is not acceptable due to high latency between the DC (example >2ms) you can consider async replication. This is done via creating a new cluster in the new DC2 and setting up one way or two way sync.

Depending on your application/site requirements, Hot/Hot, Hot/Warm, etc, query time requirement, rpo/rto requirement, you can adopt different deployment topologies.

Related