i'm playing with the 'storm-control broadcast' feature on C4K [12.2(50)SG] and C6K [12.2(33)SXH1]
i've tried three variants of the config line:
storm-control broadcast level 1.00
storm-control broadcast level 1.10
storm-control broadcast level 5.50
and used sniffers, on both sides of the 'storm-controlled' port, to examine behavior. the feature seems to work as i would have predicted: i can generate a spike of broadcast traffic (by inserting a mini-switch with a loop of cable stuck in it), but after a second or so, Catalyst shuts off in-bound traffic, and the rest of the broadcast domain returns to normal. the 'storm-controlled' port drops all in-bound traffic. after ~30 seconds, the 'storm-controlled' port drops all out-bound traffic, too, and the end-station is entirely isolated. remove the loop and this state persists for ~30 seconds. then the port reverts to normal (i.e. passing traffic in both directions).
this all seems predictable to me. however, when i read the documentation, i think i see an error:
"If you enable broadcast traffic storm control, and broadcast traffic exceeds the level within a 1-second traffic storm control interval, traffic storm control drops all broadcast traffic until the end of the traffic storm control interval."
if i removed the word 'broadcast' from "drops all broadcast traffic", then i would agree with the documentation. but as it stands, i would claim that the documentation does not accurately describe how this feature functions.
has anyone else experimented with this feature? would you agree that the feature functions as i describe it? or as the documentation describes it?
your experiment is interesting but I have a doubt: these 30 seconds timers you see are very close to STP timers for listening to learning and learning to forwarding states (15 seconds each)
What type of STP is running on your switches ?
the experiment you did involves some STP topology change with the insertion of the "miniswitch".
About the documentation I have some doubts too because it describes the feature as working on inbound traffic only but after bridging loops I saw several output drops on interfaces enabled for storm control.
I think the best way to test the feature would be to use a traffic generator so that you can send broadcast traffic at a given level with one port and to receive frames on another port.
In this way you could be able to verify if the feature description about the way traffic is dropped is correct or not.
Ideally the traffic generator should be able to send traffic at two different levels one over the threshold and one less the threshold.