From: blodwick (blodwick@columbus.rr.com)
Date: Sun Mar 05 2006 - 16:51:05 GMT-3
I'll take a shot at this one...
If the flaps occurred in rapid succession (under 5 seconds between the
first and last flap), I believe the correct answer would be:
bgp dampening 1 2250 3000 4
BTW... I really appreciate the question. It made me dig into the answer
and now I probably won't forget it :)
~~~~~~~~~~~~~~ Explaination ~~~~~~~~~~~~~~~~~
We know the penalty dampening will apply each time a route flaps = 1000
Penalties are cumulative. The penalty for the route is stored in the BGP
routing table until the penalty exceeds the suppress limit. Once the
suppress limit is exceeded the routes goes into dampened state.
Therefore if we want the route to be dampened when the link flaps 3
times we can figure out the suppress limit needs to = 3000, but this
only works if the flaps occur in rapid succession (under 5 seconds
between the first and last flap).
For the next part we need to know the exact definition of two
parameters.
Half-life Once the route has been assigned a penalty, the penalty is
decreased by half after the half-life period (which is 15 minutes by
default). The process of reducing the penalty happens every 5 seconds.
Reuse limit As the penalty for a flapping route decreases and falls
below this reuse limit, the route is unsuppressed. That is, the route is
added back to the BGP table and once again used for forwarding. The
default reuse limit is 750. The process of unsuppressing routes occurs
at 10-second increments. Every 10 seconds, the router finds out which
routes are now unsuppressed and advertises them to the world.
Therefore if we set the half-life to 1 (listed in minutes, and 1 is the
lowest allowed value), and the flaps occurred in rapid succession as
described above (under 5 seconds between the first and last flap) we
would be at a penalty value of 2250, 30 seconds after the route was
dampened. Therefore by setting the reuse limit to 2250 with the
half-life limit to 1 we would be undampened 30 seconds later. Again this
would work only if the flaps occurred in rapid succession.
Rack1R1(config-router)#bgp dampening ?
<1-45> Half-life time for the penalty
route-map Route-map to specify criteria for dampening
<cr>
Also good to remember:
Remember you cannot trigger dampening by resetting the peer connection
to test dampening.
No penalty is applied to a BGP peer reset when route dampening is
enabled. Although the reset withdraws the route, no penalty is applied
in this instance, even if route flap dampening is enabled.
~~~~~~~~~~~~~ Further Examination ~~~~~~~~~~~~~~~~~~
Further examination of why the flaps would have to occur in rapid
succession (under 5 seconds between the first and last flap) for this
answer to accurately fulfill the requirement:
I don't know if we can really honestly derive constant values. The
problem with the requirement is that you are not taking into account the
amount of time between route flaps. Let's makeup some examples...
Let's use the same dampening parameters: bgp dampening 1 2250 3000 4
Example #1
Rapid succession ( < 5 seconds between the first and last flap):
00:01:01 route 10.10.0.0/24 flap 1
00:01:02 route 10.10.0.0/24 flap 2
00:01:03 route 10.10.0.0/24 flap 3
At time 00:01:03 we would dampen the flapping route to 10.10.0.0/24.
Since we set the half-life to "1" we know that at 00:02:01 the first
penalty will = 500. Then at 00:02:02 the second will = 500 and the same
will happen to the third at 00:02:03. We also know this doesn't just
happen in one big 500 point decrement chunk, because remember the Cisco
documentation says " The process of reducing the penalty happens every 5
seconds." Also it is important to note that within the definition of the
reuse limit we learn that it will only poll for routes to readvertise
every 10 seconds (since in this question we fall on a multiple of 10
we're safe). So if we know that in 1 minute the penalty will be reduced
by 500, and we know there are 60 seconds in a minute we know that every
5 seconds the penalty for this configuration will be reduced by 8.33333
repeating (500/60 = 8.33333 repeating). Therefore in 30 seconds from the
time the penalty was applied we will have reduced each penalty by 250.
Since I set the reuse limit to 2250 (3000-250-250-250 = 2250) in my
example, as long as there were not any more flaps after the third flap,
this route would come out of the dampened state at time 00:01:33 which
is exactly 30 seconds after the third flap occurred. Also since the
route was not dampened until the third flap occurred this seems to fit
the requirements nicely.
Example #2
Non-rapid succession ( > 5 seconds between the first and last flap):
00:01:01 route 10.10.0.0/24 flap 1
00:01:16 route 10.10.0.0/24 flap 2
00:01:46 route 10.10.0.0/24 flap 3
00:01:47 route 10.10.0.0/24 flap 4
Now it should be apparent that this one is more of a moving target.
By the time flap 2 occurred flap 1 had already decreased its penalty by
125. So at time 00:01:16 the cumulative penalty on this route is 1875.
Then by the time the third flap occurred the first flap had decreased
it's penalty by 375, and the second flap had decreased it's penalty by
250. The cumulative penalty would now be 1000 + (1000 - 375) + (1000 -
250) = 2375 and would not be enough to trigger a dampening event.
Therefore we would need a forth even to occur to go over 3000. At time
00:01:47 our cumulative penalty would be 3375 and therefore in this
example it would have taken 4 flaps to dampen this route even though the
same configuration was used. It should also be obvious that this is
going to skew the time until the route will be valid for reuse. Now we
can create an equation and just solve for x.
2250 = 3375 - [(500/60 * 4) * x]
x = 33.75
So for this one we know in 33.75 seconds this route would be valid for
reuse, but since the reuse setting only polls every 10 seconds this
route really would not be valid for reuse until 40 seconds had passed
from the last flap event. Therefore in this example it would've taken 4
flaps to trigger a dampening event and it would've stayed dampened for
40 seconds even though the configuration was the same.
Brian L.
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
sameer.azam@gmail.com
Sent: Sunday, March 05, 2006 6:41 AM
To: ccielab@groupstudy.com
Subject: Dampening on the interface
Hi,
Is there is a Way to calculate the value of DAMPENING command, such as
Half Life
Reuse Limit
suppress-threshold
max-suppress-time
Ok Wt i know from the command reference is that
Max-suppress-time = 4 x Half Life
But how to answer the following Question:
Dampen the route when the link flaps 3 times then suppress the
route for 30 Sec.
Regards
This archive was generated by hypermail 2.1.4 : Sat Apr 01 2006 - 10:07:38 GMT-3