• Stop being a LURKER - join our dealer community and get involved. Sign up and start a conversation.

Reply to thread

It's a partial solution, but every VM is hosted on a host machine. If the host machine has a fault or a kernel failure, all your VMs go down at once either way. Live redundancies is the only way to truly avoid this issue, which requires separate bare metal servers, a load balancer or a docker/kubernetes build (which is what we use).

The Dealer.com one also sounds like a bad code deployment, so once again servers and VMs wouldn't stop that at all.

The solution for this is better QA process, which is incredibly complex when you have a product this widespread.