This is unreleased documentation for SUSE® Virtualization v1.8 (Dev).

Cluster Architecture and Failure Domains

SUSE Virtualization combines Kubernetes (RKE2), KubeVirt, and SUSE Storage on cluster nodes. Each layer has independent failure domains.

Failure Domain	Self-Healing Mechanism	Tolerated Failures	Incident Requiring Action
Kubernetes	etcd quorum	1/3 management nodes	Failure of 2/3 management nodes
kube-vip	ARP failover	Node switch in 30 seconds or less	VIP IP conflict
SUSE Storage	Replicas (automatic rebuilding)	1/3 replicas (default replica count is 3)	Loss of all replicas
Virtual machines	Live migration	Non-pinned virtual machines	Failure of virtual machines pinned to a specific host via affinity rules or device allocation
SUSE Rancher Prime	None (external to SUSE Virtualization)	1/3 nodes (high availability setup)	Immediate failure

Failure Domain

Self-Healing Mechanism

Tolerated Failures

Incident Requiring Action

Kubernetes

etcd quorum

1/3 management nodes

Failure of 2/3 management nodes

kube-vip

ARP failover

Node switch in 30 seconds or less

VIP IP conflict

SUSE Storage

Replicas (automatic rebuilding)

1/3 replicas (default replica count is 3)

Loss of all replicas

Virtual machines

Live migration

Non-pinned virtual machines

Failure of virtual machines pinned to a specific host via affinity rules or device allocation

SUSE Rancher Prime

None (external to SUSE Virtualization)

1/3 nodes (high availability setup)

Immediate failure

SUSE Virtualization uses StorageClasses to describe how SUSE Storage must provision volumes. Each StorageClass includes a parameter that defines the number of replicas per volume. The default replica count is 3, but you can modify this value to match your cluster’s requirements.