To achieve scaling of VRRP VR instances above 511 to as high as 2,048, selected VRRP VRs must be grouped as instructed by the user. Groups involving the same set of VRRP VRs are configured in all VRRP routers. A single chosen VR out of the group (primary VR) transmits/receives VRRP advertisements at the configured interval. Other VRs (secondary VRs) in the group transmit/receive at a much slower rate. This requires a much fewer number of VRRP advertisements exchanged, and reduces CPU resource usage, increasing scaling. All VRs in a group assume same VRRP state in a given router.
VRRP Fabric Routing needs to be enabled on all member VRs of a group in participating VRRP routers. This does not have any effect when a VR (or group of VRs) is master; but, when a higher priority master becomes available, this VR transitions to Fabric Routing-enabled backup, instead of regular backup. Since the Fabric Routing feature enables VRRP backup routers to load share by Layer 3 forwarding, and conforms to RFC 5798, this mode is preferred over using a regular backup router. However, if you want the regular master-backup configuration, you can use a separate VRID that is not used by VRs in any group.