Root Bridge Placement

[David Ruess]

I am having a bit of a contemplation. Most things I read say to put a root bridge (per VLAN) as the Core Switch. I know most, if not all, the configurations about spanning-tree and root bridges and the election process. My question is would it be better to put a VLAN's root bridge closer to where the actual computers need to go. For example, if my company hosts a First Person Shooter fight and all they need to do is connect to a "host" machine on lets say a Distribution layer switch without ever crossing to the core, would it make sense to put the root bridge for that VLAN on the "Distribution" switch? Also I am confused about the concpt of a root bridge. Does all traffic "HAVE" to pass through the root bridge to go to its destination or is it just another mechanism that is implemented to better network performance? And if so how?

Any enlightenment would be a tremendous help.

[goatnetworking]

Traffic will follow the path to the root bridge to get to it's destination. If there is an alternate path to the destination, possibly directly to another switch, that path will be in BLOCKING to prevent a loop.

Where to place the root bridge?

It's depends, but follow these ideas

1. The root needs to hold all of the MAC addresses and powerful enough to keep up with the changes

     - You don't want a small switch to be the root.

2. You want the root to be in a central part of the network, to provide the best possible data path.

3. If you built your network well, your client to client traffic should be able to travel to the Root and then to the client on another switch with a delay that is NOT noticable.

I hope this helps.

Scape

[David Ruess]

Ok when you say you want the root to be a central part of the network, is that the network as a whole or the center of where the packets "ARE" being sent. Like I said for this scenario the computers dont even touch the core for this exercise its all internal to about 4 switches, so can and should the root bridge for this VLAN be a High-End Layer 3 switch in the same COM closet, instead of the core?

[Edwin Summers]

Traffic is not required to pass through the root bridge.  Even inter-switch traffic in the same STP domain is not required to pass through the root switch.  Traffic still follows the basic rule of switches: it's switched to the port that the destination MAC address is known on (based on learning from frame source MACs).

Though extremely simplified, it may be easier to think of STP convergence as one operation, then normal frame switching.  When the network first comes online, STP does it's thing by electing a root bridge, figuring path costs, and finally converging by blocking some ports involved in redundant loops.  Once converged, assuming everything happened correctly, you can simply think of the network as a bunch of connected switches, ignoring any blocking ports.  (Traffic will not be forwarded/received on a blocked port.)

From there, just follow normal frame switching rules.  For a frame coming into a switch, the switch will review the destination MAC and switch it to the appropriate port.  If the destination MAC is unknown, it is flooded out all ports except the origin.

You can draw up a couple of sample scenarios.  They don't have to be precise, just jot down a few different switch topologies and interconnections, and cross out or erase the ports that would be blocking.  Now follow a couple of sample frames from origin to destination.  You'll see that not all would flow through the root switch.  However, due to blocking and based on switch location in the topology, there *may* be instances where frames would need to cross the entire path through the root switch to reach their destination.

STP isn't perfect, and can result in some ineffiicent traffic paths, and that is why good engineering practices are used (such as selecting the root bridge based on your network's requirements, and limiting the STP domain) to make the configuration as efficient as possible for your design.

Hope this helps.

Ed

[Joseph W. Doherty · Accepted solution]

Disclaimer

The    Author of this posting offers the information contained within this    posting without consideration and with the reader's understanding that    there's no implied or expressed suitability or fitness for any  purpose.   Information provided is for informational purposes only and  should not   be construed as rendering professional advice of any kind.  Usage of  this  posting's information is solely at reader's own risk.

Liability Disclaimer

In    no event shall Author be liable for any damages whatsoever  (including,   without limitation, damages for loss of use, data or  profit) arising  out  of the use or inability to use the posting's  information even if  Author  has been advised of the possibility of such  damage.

Posting

Root switch placement is optimally determined by trying to achieve the shortest number of hops for most of your traffic.

Suppose you had 5 switches connected in a ring.  Let's assume most of your traffic was between switches 1 and 5.  If you made switch 3 the root bridge, normally the link between 1 and 5 would block, so traffic between switches 1 and 5 would have to transit switches 2, 3 and 4.  If, however, you made either switch 1 or 5 the root, then the traffic between these switches could use the directly connecting link.

In many topologies, the "core" is the central node in a hiearchal topology.  As such, it's often the "best" root bridge.

With VLANs, and per-VLAN spanning tree, you can have different logical topologies.  So, again with a physical ring, switch 3 might be the best root for one VLAN and switch 1 the best root for another VLAN.

When your underlying physical topology is partially meshed, not only can you design the optimal topology for certain transit traffic in "normal" (i.e. everything operating) but best alternative topology for device failures.  This can get quite complex, which is also why dual hierarchal physical topologies are used with a primary and alternate root being on core bridges.

[goatnetworking]

David

From what you have stated, it sounds like it makes sense to place the root bridge close to the end users for that VLAN to keep the traffic local.

I would do that

Scape

[anuarus]

i can see that this is resolved, usually you place the root switch in the same switch that is doing the routing for that vlan (AKA where the SVI is) the root is not only a mechanism to find the closes point to the IP gateway, but also to define the path of traffic the most efficient way, if you do have a ring type topology then it comes into play, if your network is (spoke and hub) then the hub should be the root, but this varies, since you can have multiple gateways distributed thru out the network (BAD design)
in short closest to the SVI and only when mulitple paths exists.