Solved: STP Optimise BPDU Transmission?

pgalligan · ‎03-05-2006

Can anyone tell me what this command does. The default for our switches is for it to be disabled, ie. "no spanning-tree optimise bpdu transmission". I have checked the documentation for the relevant switch models and can't find it in the command references or configuration guides.

thanks

Francois Tallet · ‎03-06-2006

That's really a subtle IOS architectural optimization so I don't want to enter into the details, but basically:

-Without optimized transmission, BPDUs are sent independently one by one from the STP process.

-When optimized BPDU transmission is used, the BPDUs for a given vlan are prepared first (they all look the same in fact) and then sent in a row at interrupt level (with no further CPU interruption). The main advantage is that the same packet buffer is allocated once and recycled for all the BPDUs, instead of being allocated and freed for each and every BPDU.

Regards,

Francois

View solution in original post

arvindchari · ‎03-05-2006

Usually I refer to

http://www.boerland.com/dotu/

for undocumented command references.

This one does not seem to be listed there!

Francois Tallet · ‎03-05-2006

The optimize bpdu transmission is a mechanism that allows BPDUs to be sent in one shot at interrupt level. It is supposed to relieve the STP process in PVST mode. It's not a fundamental feature and should not have been exposed to the user. The fact that this is disabled (and thus that it shows up in the configuration) is a bug. So basically, don't worry too much about that. When you have a release that allows to enable it again, you can do it and forget about it.

Regards,

Francois

pgalligan · ‎03-06-2006

When you say "sent in one shot" what exactly do you mean?

Thanks!

Francois Tallet · ‎03-06-2006

That's really a subtle IOS architectural optimization so I don't want to enter into the details, but basically:

-Without optimized transmission, BPDUs are sent independently one by one from the STP process.

-When optimized BPDU transmission is used, the BPDUs for a given vlan are prepared first (they all look the same in fact) and then sent in a row at interrupt level (with no further CPU interruption). The main advantage is that the same packet buffer is allocated once and recycled for all the BPDUs, instead of being allocated and freed for each and every BPDU.

Regards,

Francois

pgalligan · ‎03-06-2006

OK, thanks.

DAVE GENTON · ‎03-28-2006

Perfect, thanks