[c-nsp] Too much HSRP traffic - how to limit?

[Phil Mayers]

Gert Doering wrote:
> Hi,
> 
> On Tue, Feb 27, 2007 at 08:10:35PM +0000, Phil Mayers wrote:
>> Combined with the HSRP "follow" groups also in 12.4T 
> 
> What's that?
> 
> gert


Sorry - the correct feature name is "HSRP multiple group optimisation" 
but that name doesn't give the complete picture. Basically, standbys on 
one interface can follow another:

int f0/0.1
  encap dot1q 1
  standby ip blah
  standby name MASTER

int f0/0.2
  encap dot1q 2
  standby ip foo
  standby follow MASTER

The client standbys emit slow periodic messages to refresh the arp and 
fdb tables of clients on that subnet, but do not need to run millisecond 
timers (or lots of millisecond BFD sessions).

One hopes that if/when it makes it into the switching platforms you'll 
be able to use it on SVIs, and there isn't some silly limitation about 
all the underlying interfaces having to be the same physical int.

It would be very handy for e.g. 20 subnets in a building but only one 
topology.

See:

http://tinyurl.com/2vj3rm

On a related note I've been asking Cisco for a while now for "Per-VLAN 
STP follow" which would do much the same thing, and is present on both 
Foundry and Extreme:

spanning-tree vlan 2-10 follow 1

...which to my mind is FAR more useful than MST, given that you can only 
run one MST process on a Cisco. Combined with:

spanning-tree vlan 1 pdu-version 8021d

...for interop with single STP switches I could actually make use of STP.