Lessons › Servers & virtualisation
Storage, clusters and keeping VMs running
▶ Start the interactive lesson
At 2 a.m. a host's motherboard dies. Twelve VMs were running on it. By 2:04 they are all running again, on the other three hosts. Nobody was called. The next morning your manager asks: "Do you know why that worked? And why it wouldn't have worked last year?"
Shared storage
If a VM's disk lives on the host's own drives, the VM is stuck on that host. If the host dies, so does the VM.
Clusters, live migration and HA
Hosts that share storage can be joined into a cluster.
Plan for N+1
If four hosts are each 90% full, there is nowhere for a failed host's VMs to go. Size the cluster so it can lose one host and still run everything: N+1.
Key points
- Shared storage lets any host run any VM.
- Live migration moves running VMs with no downtime; HA restarts them after a host fails (minutes, not zero).
- Size clusters N+1, and keep backups: HA doesn't protect your data.
Words you'll learn
Shared storage · SAN (iSCSI / Fibre Channel) · Hyperconverged (HCI) · Cluster · Live migration · High availability (HA) · N+1