The network design for the cluster itself is essentially the same. The three nodes physically connect to a separate storage network, a separate heartbeat/live migration network and a separate virtual machine/front-end network.
I’ve redesigned my entire network, of which this cluster is just one part, to be physical-cabling-connection independent and fully VLAN segmented including authentication, routing, firewalling, etc. The ins and outs of this design may be a series of its own someday.
Anyways, this setup allows for very flexible placement of VM’s in any VLAN to enable isolated tests, integrate a VM in my internal production network or easily expose a VM to the Internet without changing any physical stuff.
**The IP addresses and notes in the diagram need to be updated**
1. the BLUE network: Virtual Machine(s) front-end
This network is part of my C2 VLAN and is the primary placement for VM’s. The cabling is blue.
172.16.232.x /26 (62 hosts – subnet 192)
2. the YELLOW network: TTG Storage Network
I’ve separated the storage network which is the only exception to the rule and it runs on a separate physical network. Each cluster node will receive an address in this range to be able to boot from NAS, and to provide access to the Cluster Shared Volume. The cabling is yellow.
172.16.12.x /28 (14 hosts – subnet 240)
3. the GREEN network: TTG Heartbeat / Live Migration / Cluster Nodes Network
This network connects each cluster node to the C4 VLAN to enable Live Migration between the nodes and to provide the cluster heartbeat. The cabling is green.
172.16.211.x /26 (62 hosts – subnet 192)
Cisco Nexus 1000v
The Cisco Nexus 1000v is now available for the Hyper-V Server 2012 platform. I’ll see how to incorporate these new features in the network design in a future post.