[f-nsp] Trunk / LAG and spanning-tree costs

[Nick Cutting]

Thank you,

Great find - I could not find this in the documentation.

I think I'll manually change the link of the "boss" link in the LAG, although now I'm moving the root, it is not strictly needed in my case anymore.

-----Original Message-----
From: foundry-nsp [mailto:foundry-nsp-bounces at puck.nether.net] On Behalf Of Franz Georg Köhler
Sent: Wednesday, June 1, 2016 4:39 AM
To: foundry-nsp at puck.nether.net
Subject: Re: [f-nsp] Trunk / LAG and spanning-tree costs

Am 27.05.16 um 02:59 schrieb ncutting at edgetg.com (Nick Cutting):
> How does spanning tree treat trunk links a.k.a Etherchannel/Aggreagated links?
>
> I would have thought that the pseudo 20 gig bandwidth would lower the Cost of the link ?

No, it is the same cost.

This is from the manual:
http://www.brocade.com/content/html/en/configuration-guide/NI_05800a_SWITCHING/GUID-052E20B0-FF5A-41F5-BB48-A2370F0D5C5F.html

STP in a LAG
The STP standard indicates that by default the path cost is determined by link speed. For a1 G port the path cost is 4 and for 10G port the path cost is 2. However, if a LAG is made consisting of n 1G ports, where n is less than 10, the path cost remains as 4. The standard does not indicate pathcost explicitly for LAG interfaces or for bandwidths between standard port bandwidth values, (for example, between 1G and 10G). Therefore, during STP deployment you may find that though a LAG has greater bandwidth, its in blocking/discarding state as its pathcost is the same as any 1G link and the portIndex of 1G port is lower, making the LAG go into a blocking/discarding state. This behavior is not restricted to 1G or 10G link speed but span across different link speeds. The same behavior also holds TRUE for RSTP deployments.




Best regards,

Franz Georg Köhler

_______________________________________________
foundry-nsp mailing list
foundry-nsp at puck.nether.net
http://puck.nether.net/mailman/listinfo/foundry-nsp