Multimaster database cluster with writes to all masters even during network separation

Viewed 21

I have multiple locations with (sometimes) bad Internet connections, and in every location I am gathering sensor data.

Right now I am writing to a central database, but if there is a connection failure I lose all data. It might take several weeks to restore the connection.

I thought about caching the data at every location, wait for the connection and then write everything, but that's quite a lot of programming and a lot of possible bugs.

Is there a prebuilt solution for this problem? A db-cluster which accepts reads even if the cluster is separated and then synchronizes again if the connection is back online?

I would prefer a time series database, but a sql-db would do too.

Thanks in advance

1 Answers

You can use any distributed db such as cockroach or yugabyte.

Upon failures a new leader will be elected and all writes will seamlessly be forwarded to it

As long as you have more than n+1 /2 nodes your cluster should be good for both reads and writes

Related