[c-nsp] VoIP without QoS

[John Osmon]

It's certainly not as rigorous as one might hope, but it does explain
why common VOIP applications running over 802.11b start to fail when
around the time the 7th conversation is added to the mix.  Which is
quite consistant with empirical evidence I've collected, and anecdotal I've
had from others.

My main point is that if the transport is prone to congestion with a
(relatively) small number of pps, it isn't good candidate for VOIP
communications.  The commonly found wireless schemes in use today fall
into this category of transport.  We'll see things get better as time
goes on, but the current crop of wi-fi gear simply isn't suited for VOIP
except in very spcific situations.

I almost feel sorry for the WISPs that have built out neworks with 
802.11b backahul links.  I've *seen* some of them staring at MRTG
graphs saying "but there's less than 1 Mbps on that link!  We can't
be the cause of your VOIP problems..."  As we all know -- the pps 
numbers are more imporportant than the bandwidth nubmers in most 
situations.

This has drifted too far from the 'c' in 'cnsp' -- I'm going to go back
to lurking...



On Sat, May 26, 2007 at 11:51:42AM -0500, Frank Bulk wrote:
> In the paper: "To generate a 64 Kbps flow, 64 bytes packets were sent every
> 8 milliseconds."
> 
> This is not how the G.711 codec works...it's normally a 20 msec
> packetization rate resulting in ~220 byte frames (sans 802.11 headers, I
> believe).  I've seen too much undergraduate start with faulty assumptions
> such that the results are bogus.  The only thing these guys have proven is
> that lots of small packets cause latency and jitter when the link becomes
> saturated.  No surprise there.
> 
> Regards,
> 
> Frank
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of John Osmon
> Sent: Tuesday, May 22, 2007 11:21 PM
> To: cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] VoIP without QoS
> 
> On Tue, May 22, 2007 at 01:47:15PM -0500, Dan wrote:
> > We have a voip system we have been running in our department now for
> > about a year.  Only 12 phones, connected through various wireless links
> > with throughput of up to 40mbit.  Speed is definitely not an issue for
> > us, but we notice glitches with the quality on an ongoing basis.  We are
> > currently implementing qos and are wondering what is the best way?
> 
> Most common wireless solutions don't really like a lot of small packets.
> They tend to have too much overhead in the protocols, and you if you're
> pushing small packets, you hit a pps limit long before anything comes 
> close to using up the bandwidth you think is available.
> 
> While trying to research why VOIP over wireless networks, I ran into
> this paper:
>    http://www.it.uc3m.es/~acuevasr/publicaciones/LCS06.pdf
> 
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>