Hello,
We have two data centres and are considering HNV to stretch our VM network across sites, in order to avoid re-addressing VMs on failover. The HNV gateway would be in forwarding mode with appropriate routes configured on an L3 switch. The data centres only have layer 3 connectivity so we can't stretch VLANs, but we can route PA subnets between sites. HNV would also permit us to leverage Azure for workloads in future. The data centres are considered primary and backup, and each has an internet connection and link to our MPLS
My question is around HNV gateway redundancy. Should the primary data centre fail, and that's where the HNV gateway is hosted, how can we restore connectivity to the HNV network?
1) Multiple HNV Gateway clusters (one per site)? Does SCVMM then use these in a round-robin fashion until one fails? Can multiple network services on a VM network be configured active/passive? Is Hyper-V/SCVMM intelligent enough to use a gateway on the same PA subnet as the hosts?
2) Multi-subnet cluster for the HNV gateway guest cluster? (e.g four nodes, two in one subnet, two in another). Will SCVMM have problems when the front-end IP is changed by the cluster?
I haven't seen much guidance around HNV gateways in a multi-site private cloud so any input would be appreciated.
Thanks