Is this a technical issue or a manpower issue? This steers us strongly toward Galera. Why?
Our operator is specifically designed for PXC/Galera because that makes the most sense in a K8S environment. If you have a single MySQL server, you have a single point of failure. Even if K8S restarts your MySQL pod, you still have downtime. By using PXC, if a pod fails, you have immediate failover to a survivor pod because it is part of your cluster.
Hello @1d2u3c , let me add something here.
Percona has the Distribution for MySQL. In there we have Percona XtraDB Cluster (PXC) and Percona Server for MySQL (PS).
PXC historically was the tool to deliver ready-to-use HA MySQL cluster with synchronous replication based on Galera.
Due to ephemeral nature of Kubernetes we decided to provide sync replication for MySQL in our Operator, hence the choice of PXC.
Right now we see that Group Replication (GR) also getting more and more mature, so we have the plan to have an Operator based on Percona Server for MySQL with async (regular replication) and sync replication based on GR.
BTR, do you have any issues steering toward Galera?