cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1533
Views
0
Helpful
1
Replies

Switching redundancy for ESXi VMHosts scenario

ziggitarrius
Level 1
Level 1

I have the following devices :

1 VM Host

2 Layer 3 switches

I would like to provide full redundancy for all vlans being used by VM Guests on the VM Host as well as the management vlan being used by the VM Host.

I have created two LACP etherchannel connections on the VM Host. Each etherchannel from the host consists of  4 ports spanning a single NIC. One etherchannel connection goes to a trunked etherchannel connection on switch 1, and the other etherchannel connection goes to a trunked etherchannel connection on switch 2.

Switch 1 and switch 2 have an etherchannel connection between them that carries all of the vlans in the topology.

Vlan 2 is the managment vlan. Vlans 3, 4, and 5 are vlans that VM guest systems will be using for normal data traffic.

I intend to use switch 1 as the VRRP active router and spanning-tree root bridge for vlans 2 and 3.

I intend to use switch 2 as the VRRP active router and spanning-tree root bridge for vlans 4 and 5.

The spanning-tree configuration is using multiple spanning-tree with two instances. Instance 1 has vlans 2 and 3 associated and Instance 2 has vlans 4 and 5 associated.

I would like to have this topology be fault tolerant to the point where if one of the etherchannel links between the host and one of the switches goes down, (for example, if switch 1 was powered off) traffic will be automatically redirected through the other functional link.

I believe that my VRRP configuration would allow for a fairly quick failover of layer 3 services, but I am not certain that my design will be functional at a layer 2 level.

What I am uncertain about is how spanning-tree will converge. I am assuming that the virtual switch on the VM host will not be forwarding any BPDUs being sent by either switch.

Would either of the links connecting to the host be considered a redundant link by either switch?

Would the link between switch 2 and the host be inactive for all vlans in MST instance 1 during normal operation?

Conversely, would the link between switch 1 and the host be inactive for all vlans in MST instance 2 during normal operation?

Would all links remain active for ALL vlans? Would this mean that some traffic may travel through switch 2 to reach switch 1 instead of going directly to switch 1?

I have attached a very basic visio diagram photo to help illustrate the topology.

I really appreciate any insight that could be given on how I should expect this design to function. Thank you very much in advance!

1 Reply 1

ziggitarrius
Level 1
Level 1

Adding all information I find as this may conceivably help somone else later.

ESXi does not bridge between uplinks so no BPDUs would ever cross from switch 1 -> through the host -> then to switch 2 or vice versa.

Would either of the links connecting to the host be considered a redundant link by either switch? No, no bridging being performed between any uplinks

Would the link between switch 2 and the host be inactive for all vlans in MST instance 1 during normal operation? No, all links would be active simultaneously

Conversely, would the link between switch 1 and the host be inactive for all vlans in MST instance 2 during normal operation? No, see the answer above this

Would all links remain active for ALL vlans? YES  Would this mean that some traffic may travel through switch 2 to reach switch 1 instead of going directly to switch 1? Not sure yet...

Review Cisco Networking products for a $25 gift card