Standalone-HA vs. Domain Clustered Mode

I try to move Keycloak to the production level, I read that it’s not recommended to use the Standalone mode in the production environment. So, I have two options, Standalone-HA and Domain Clustered Mode. I read about both of them but it’s really tough to choose one of them because the documentation doesn’t say a lot. So, what do you see and what is your recommendation?

1 Like

Thank you so much for your response.
I just want to make sure of something and please tell me if I’m right.
I need to deploy Keycloak on Google Cloud Kubernetes with a central failover database. In this case, Standalone HA is enough and I won’t need to do configurations manually because the new instances in the cluster are created automatically from a Keycloak docker image. Am I right?
If this is true, how I can synchronize caching in the cluster for the running sessions?

Yes, in your case standalone HA is enough.

The cache in a cluster will sync automatically, if nodes find each other. When using Kubernetes, have a look into “kub_ping” or “kubeping” for wildfly/infinispan cluster.
Alternatively, also JDBC_PING should be possible. See also Keycloak - Blog - Keycloak Cluster Setup and Keycloak - Blog - Keycloak and JDBC Ping

2 Likes