The Routing Over Large Clouds Mailing List Archive by date

[gardo@vnet.ibm.com]

Jim,

>> As proposed earlier, why not add an "acknowledgement request" flag in
>> the Purge packet.  There are environments and configurations where
>> acknowledgments are very useful, and I agree with you that there are
>> other environments/configurations where it is actually problematic.
>You are stating a case for exactly one of the scenarios for which
>the vendor private extension was created.  You categorically cannot add
>mandatory protocol mechanisms that cause, by your own admission, such
>"problematic" scenarios.

Ignoring the Purge is also a problem.
Let's move the purge protocol one way or the other, instead of
putting it halfway between required and not-required.  Move the purge
protocol to the extensions part of NHRP; otherwise, keep the ack
or add an "ack request" flag to the Purge request...

>> After the holding time expires, the server will stop sending Purges.
>> How many re-transmissions do you think a server will send before the
>> holding time expires?
>
>Based on your previous argument, "a long holding timer", the answer is
>quite a few just when extra traffic is highly undesirable!

Some implementations would use a constant maximum number of retries,
and IF WE HAD a purge mask the number of Purges would be reduced.
Is there a consensus on the purge mask?  I didn't see a response
from you to Andrew.

Have a nice day!
-- Russell