Solved: What is The Reason For the Root

mrashby · ‎12-26-2007

All,

I have a VLAN config question. I was reading this Cisco book about VLANs, VTP, & STP and I saw a configuration I didn't understand. Here it is, There were three switches Main/server switch and two distribution/client switches. Well these swtiches had Vlans 10,11, and 13 on them. Now on the main switch they configured that switch to be the root for VLan 1. On switch two they configured that switch to be the root for vlan 11. On the third switch they configured that one to be the root for Vlan 13. My question is why would you configure a switch to be the root switch of a particular VLAN? What purpose does this serve?

Jon Marshall · ‎12-26-2007

Again this is generally speaking but not not really unless you wanted very specific control over your layer 2 paths.

A common setup is to have a pair of core switches and you make switch1 the root for all odd vlans and the secondary for all even vlans, and switch 2 the root for all even vlans and the secondary for all odds.

If you have a layer 2 access-layer connecting to a pair of distribution switches that do the L3 vlan routing then these would be the switches to set as root and secondary.

Cisco's 3 tier design - core/distro/access is not always that clear cut in terms of physical network infrastructure. Quite often the same pair of switches act as the distro and core.

Jon

View solution in original post

ankbhasi · ‎12-26-2007

Hi Mario,

This is the advantage of PVST over CST. In CST you have single spanning tree instance for all vlan and in PVST you have per vlan spanning tree instance and different switches can be root for different vlans depending upon network topology.

Basically this helps in achieving optimal paths.

HTH

Ankur

*Pls rate all helpfull post

Jon Marshall · ‎12-26-2007

Hi

Assuming you are referring to spanning tree root.

Lets assume you have 2 core switches and 4 access-layer switches with vlans 10 -> 15 across all switches. The access-layer switches connect back to the core switches.

Now if, you don't explicitly set the spanning-tree root for these vlans any of your switches could end up being the spanning-tree root for those vlans. Assuming all switches have a default priority of 32768 then the mac-address of the switch will be the decider.

And so you may end up with one of your access-layer switches being root which could have the following issues

1) access-layer switches generally don't have as much "horsepower" as core switches. Depending on the number of vlans this could be an issue.

2) You could end up with sub-optimal layer 2 paths on your network.

Generally speaking, and it is generally, you would pick your core/distribution layer switches to be your spanning tree root and secondary depending on where your layer2/3 boundaries were in your network.

HTH

Jon

mrashby · ‎12-26-2007

So generally speaking if I had Vlans 10-30 is there any sense behind making my core switch the root for Vlans 10-20 and my distribution switch the root for Vlans 21-30. Or would it just make more sense to just make my core switch the root for all of the Vlans.

royalblues · ‎12-26-2007

If you have redundant core/distribution, it would make more sense to have one switch to be root for some vlans and the other one to become the root for the remaining vlans

HTH

Narayan

ankbhasi · ‎12-26-2007

Hi Friend,

As I said lot of things depend upon your topology. Basically having multiple root bridhes can provide load balancing and also redundancy.

You can define one core switch as root for few vlans and other core switch for other vlans and also second core switch can work as secondary root for those vlans who have core 1 as root and vice versa.

HTH