[83704] in North American Network Operators' Group
Re: Question about propagation and queuing delays
daemon@ATHENA.MIT.EDU (Robert E.Seastrom)
Mon Aug 22 15:30:07 2005
To: "Howard, W. Lee" <Lee.Howard@stanleyassociates.com>
Cc: "Richard A Steenbergen" <ras@e-gerbil.net>,
"Joe Abley" <jabley@isc.org>, "David Hagel" <david.hagel@gmail.com>,
<nanog@merit.edu>
From: Robert E.Seastrom <rs@seastrom.com>
Date: Mon, 22 Aug 2005 15:29:37 -0400
In-Reply-To: <3F05EE24A82C0B42811178EFB8820C3F268AA6@AX-S-EX-1.stanleyassociates.com> (W.
Lee Howard's message of "Mon, 22 Aug 2005 15:00:17 -0400")
Errors-To: owner-nanog@merit.edu
"Howard, W. Lee" <Lee.Howard@stanleyassociates.com> writes:
> Do carrier ISPs classify their voice traffic as Really
> Important, and everybody else's data as Best Effort? This
> isn't just selfishness, since We All Know voice is less
> tolerant of latency and jitter than TCP.
We do?
Try to keep a single-TCP-stream session over 500 mbit/sec of goodput
halfway across the country sometime. Most of us who have VoIP boxes
have 20ms or better jitter playout buffers. Ever (successfully) talk
to someone over a GSM handset when their coverage was a little
marginal? The human ear is a marvelous tool compared to a TCP timer.
Apologies if "We All Know" was supposed to be a cue to sarcasm or irony...
---Rob
