From: Erick B. (erickbe@yahoo.com)
Date: Thu Oct 17 2002 - 04:13:54 GMT-3
I've been involved in alot of ATM-FR voip stuff lately
(post-sales fix-it up stuff)... and I'm pretty sure
the answer to this is no or not recommended but wanted
to bounce it off you guys.
Does anyone know if FRF.12 (Fragmentation) can be done
on the FR side and if the carriers frame switch will
piece it back together before sending it over to ATM?
I haven't tried this, just a idle curiousity I had.
Anyway to do RTP header compression on ATM (without
doing PPP over ATM) ?
What I've been doing mainly is using LLQ on both, on
ATM PVCs - tx-ring-limit and vbr-nrt or vbr-rt
w/burst, FRTS+LLQ on frame PVCs. Working well but
wondering if I can tweak some more.
As for item #4 mentioned (diff serv) - I find it
better to match an ACL and specify the port ranges,
etc as well in addition to matching the diff serv
settings.
Erick
--- Donny MATEO <donny.mateo@sg.ca-indosuez.com>
wrote:
> There area lotsa thing that you can do in Voice
> depending on the case.
> 1. Is ip rtp header compression
> 2. LFI (link fragmentation and Interleafing) to make
> sure packet delay is not more than 10 ms
> 3. MQC (do remember CBWFQ guaranteed minimum
> bandwith and that it's only kick in during
> congestion.
> However voice is delay sensitive, so use LLQ
> instead).
> 4. Differentiated Services (DS) and Differentiated
> Services Code Point (DSCP)
> 5. Traffic shaping for Frame relay
> 6. RSVP LLQ + QOS for signaling Traffic.
>
> there are rules to this optimation, like use ip rtp
> header-compression only on slow link and when
> the number of call are small. Use LFI on link slower
> than 1.2 Mbps etc
>
> the link below might help
>
>
http://www.cisco.com/en/US/tech/tk652/tk701/technologies_white_paper09186a00800d6b73.shtml
>
>
> Donny
This archive was generated by hypermail 2.1.4 : Tue Nov 05 2002 - 08:35:49 GMT-3