[c-nsp] Frame-Relay fragmentation

Luan Nguyen luan.nguyen at mci.com
Thu Feb 17 11:30:44 EST 2005


Hi all,
Mostly on CCO, the suggestion is that if you have a link > 768K you don't
need to use frame-relay fragmentation.  My guess is because the
serialization delay for 768K would come out to be ~15ms = within the
suggested range of Cisco for voip to work.  But I read a book on QOS -
forgot the name, but written by 2 CCIEs geering toward the QOS exam I think,
and in the book, they suggest to use frame-relay fragmentation with a high
number such as 1536 or so for even highspeed link > 768K, so that you won't
have to fragment packets but still have the benefit of a dual fifo queueing.
the LLQ traffic in the service-policy will always get pick by the high
priority fifo queue.
Any comment on this?

Thanks.




More information about the cisco-nsp mailing list