From: Jeff K. (jeffbk@xxxxxxxxxxxxx)
Date: Tue May 15 2001 - 23:28:22 GMT-3
Every packet does not need to go through the root bridge. But, yes, you can
definitely have a non-optimal switching path if you allow the network to
pick the root for your vlan. Like Ron said earlier, the root bridge is
basically a reference point from which all the other switches will select a
path in order to prevent bridge loops. In a layer 2 network that is set up
as a hub-spoke, for example, then all traffic destined for another switch
will pass through the root - since your root should be the core switch.
But, remember, the switch forwards packets based on the CAM table. This is
definitely affected by root bridge location, since that determines
forwarding ports. But, still, a packet doesn't need to hit the root. For
example. say switch A is root, switch B is linked to switch A, and switch C
is linked to switch B (A---B---C). A packet from a host on B needs to go to
a host on C. That packet will not touch the root bridge. The root bridge
will allow B to determine which of its redundant links to A it will use to
connect to A (and same for C's links to B). The main thing to remember with
the root bridge is that it plays a big role in network stability. You want
it to be a switch that is as stable as possible since a new root will cause
a topology change. You also want it to be a strong box. This gets back to
having a good design. When you select the best location for your root,
change priorities (or us the 'set spantr root' macro) to force this. With
equal priorities, the lowest MAC address wins, meaning an old router that is
bridging would probably become root over your nice, new switch since it has
the old, low MAC address... That could easily cause the other switches to
choose non-optimal paths. It's also a good idea to choose your secondary
root and lower it's priority as well (use the 'set spantr root secondary'
macro). A nice layer 2 design is a pair of beefy core switches with a very
high-bandwidth link between them, one root for half VLANs and the other root
for the remaining. They are each secondary roots for the other. The access
/ distribution switches are dual homed to the 2 core switches and can run
nice features like uplinkfast and backbonefast for fast reconvergence
times...
HTH (sorry if it is too long),
-Jeff
----- Original Message -----
From: "sanjay" <ccienxtyear@hotmail.com>
To: "Rob Webber" <rwebber@callisma.com>; "Rob Rummel"
<Rummel@Hawaii.RR.com>; <ccielab@groupstudy.com>
Sent: Tuesday, May 15, 2001 8:07 PM
Subject: Re: Dumb Bridging Question
> Switching and bridging sometimes result in non optimal routing of packets
> because every packet must go through the root bridge of the spanning tree.
>
>
>
>
>
> ----- Original Message -----
> From: "Rob Webber" <rwebber@callisma.com>
> To: "Rob Rummel" <Rummel@Hawaii.rr.com>; <ccielab@groupstudy.com>
> Sent: Tuesday, May 15, 2001 6:27 AM
> Subject: RE: Dumb Bridging Question
>
>
> > I don't think this is a dumb question - many people don't fully
understand
> > Spanning Tree.
> >
> > No, all traffic does not have to flow through the root. It is simply
used
> as
> > a reference point from which all bridges create a network with no loops.
> >
> > The root is the bridge with the lowest number for root priority. On
every
> > segment one bridge is the "designated" bridge. This bridge will be
> > forwarding traffic from that segment. The designated bridge is the one
> with
> > the lowest path cost to the root. The path cost is calculated by adding
> the
> > outbound costs on all segments to reach the root. In the case where two
or
> > more bridges have the same path cost to the root, the designated bridge
is
> > the one with the lowest priority.
> >
> > You can start drawing some diagrams with one root bridge and many other
> > bridges. Create some loops and start seeing which links have to be in
> > "blocking" mode to prevent loops (and how you can control which ones are
> > blocked based on path cost). It will become clear not all traffic HAS to
> > pass through the root bridge.
> >
> > Good luck - Rob.
> >
> > -----Original Message-----
> > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
> > Rob Rummel
> > Sent: Tuesday, May 15, 2001 12:35 AM
> > To: ccielab@groupstudy.com
> > Subject: Dumb Bridging Question
> >
> >
> > Cant believe I'm so far along my studies but don't understand a basic
> > fundamental principle so please bear with me.
> > When a Root bridge is elected does all the traffic in that STP go THRU
> > the route bridge or is it just for REFRENCE for STP????
> >
> > Any sufficiently advanced technology is indistinguishable from magic.
> > - Arthur C. Clarke
> > **Please read:http://www.groupstudy.com/list/posting.html
> > **Please read:http://www.groupstudy.com/list/posting.html
> **Please read:http://www.groupstudy.com/list/posting.html
**Please read:http://www.groupstudy.com/list/posting.html
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:30:42 GMT-3