This seems flawed to me - Do I understand correctly that Justin is saying I could have a cluster of 100 servers and if only the primary and the backup it is connected to die at the same time then the cluster is down and I loose data? So Artemis is NOT high availability