The World’s Largest Online Community for Developers
I am using Redis on ElastiCache for a Node application and today the node went down which means our app stopped working. It took 20 minutes for a new node to be provisioned.
From reading the documentation it seems I can set up a cluster which automatically promotes a slave to primary in case of a failure. The big gotcha seems to be you have to set your client to write to the primary node and read from the slave nodes.
This means in the case of a failure, you have to reconfigure your app to point to the newly created 'read' nodes. It also takes a few minutes for a slave to be promoted to primary.
Is there no way to set this up so if the primary fails, a slave will automatically take over for read/write operations?
I'm not storing much data in redis and low read/write operations, but it is required to run the app (live video sessions!).
If I can't have a seamless failover in redis, is there something I can use which provides this functionality? I'm hoping I don't have to move to a traditional DBMS as everything works perfectly but I need to be able to handle failure well.
Multi AZ's should automatically switch over with minimal downtime. Once you have created one of these instances, you will get an endpoint for the cluster. Amazon will point that DNS entry to the proper failover node, and handle the promotion of a slave, if the master instances dies.