RE: STP cost

From: Geert Nijs (Geert.Nijs@simac.be)
Date: Mon Feb 11 2008 - 11:36:29 ARST


Spanning tree always only blocks one side of a connection. The interface stays blocked by the receiving of BPDU's from the other side.
If i remember correctly (it has been a while, up for recertification shortly :-), STP takes the cost from the received BPDU packet, add the cost of the RECEIVING interface to the value
and advertises this back out all other interfaces (cost defined on those interfaces does not matter)

regards,
Geert
CCIE #13729

________________________________________
From: nobody@groupstudy.com [nobody@groupstudy.com] On Behalf Of thomas.rader@freesurf.ch [thomas.rader@freesurf.ch]
Sent: 08 February 2008 17:13
To: ccielab@groupstudy.com
Subject: STP cost

hi,

I thought I understood the STP cost concept until I labed a simple scenario and couldn't work out why ports were being blocked.

Here is what I did:

My scenario has an access switch (SW1) connected to two distribution switches (DS1 & DS2) via 2xgig etherchannel (stp cost 3)
The distributibution switches (DS1 & DS2) are also connected via gig trunk (stp cost 4).
DS1 is STP root.
I used the "spanning-tree cost" under ther interface or in the case of etherchannel on the port-channel.

If I change the STP cost on SW1 etherchannel connected to DS2 to 1000, I would expect that STP blocks SW1 port to DS2.
However in fact it blocks DS2 connected to SW1. Why ?

Here is how I see it:
SW1 to DS1 = cost 3
SW1 to DS1 (via DS2) = cost 1004 (1000 + 4 crossing the DS2-DS1 link)
DS2 to DS1 = cost 4
DS2 to DS1 (via SW1) = cost 6 (3 + 3)

This assumes that only the outgoing cost per hop is added to the cost. Is this correct ?

I have seen some confilicting information regarding per VLAN ("spanning-tree vlan 1-1024 cost"). The way I see it you can simply control this better per per VLAN. Right ?

Also why does STP only block one side of the connection ? I would have expected both sides of a connection to be blocked ?

thanks, thomas



This archive was generated by hypermail 2.1.4 : Sat Mar 01 2008 - 16:54:48 ARST