From: Chuck Church (cchurch@optonline.net)
Date: Fri Nov 22 2002 - 02:04:23 GMT-3
Eric,
Check CCO for their docs on fine-tuning VoIP over frame. They've got a
bunch of them. From what I remember, you want to keep your burst amount
very close to your CIR, if any burst is allowed at all. The intention is to
not have any frames dropped because the frame switch flagged them DE. Other
things like frame relay fragmentation are there as well, which guarantees a
maximum serialization delay, which you want to limit to 10 - 20 ms. Of
course RTP priority queueing works well. It can be some trail and error
finding a combo that works, but it's possible. I managed to get a VoIP
implementation working over 56 kb circuits shared with data traffic and STUN
for SNA traffic. It's really cool when you make a change, and then make a
call to see if it made it better or worse. Hope this helps.
Chuck Church
CCIE #8776, MCNE, MCSE
----- Original Message -----
From: "eric" <namaste@pacbell.net>
To: "ccie" <ccielab@groupstudy.com>
Sent: Thursday, November 21, 2002 10:07 PM
Subject: frame-relay traffic shaping
> Does anybody have some baseline numbers to use with voip across a
> frame-relay connection. I am a bit confused with the different books since
> they all give the examples but no real explanation as to why or how they
> come up with these numbers. Solie book comes up with completly diff.
> numbers.
>
> ie..cisco voice over frame-relay page 261
> frame cir 12800
> frame bc 1280 why 1%
> frame be 0 ?
> no frame mincir ?
>
> thanks
>
> ~eric
This archive was generated by hypermail 2.1.4 : Tue Dec 03 2002 - 07:23:08 GMT-3