[133998] in North American Network Operators' Group
RE: Some truth about Comcast - WikiLeaks style
daemon@ATHENA.MIT.EDU (Rettke, Brian)
Tue Dec 21 11:26:59 2010
From: "Rettke, Brian" <Brian.Rettke@cableone.biz>
To: Lamar Owen <lowen@pari.edu>, "nanog@nanog.org" <nanog@nanog.org>
Date: Tue, 21 Dec 2010 09:26:48 -0700
In-Reply-To: <201012210808.52662.lowen@pari.edu>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
Sincerely,
Brian A . Rettke
RHCT, CCDP, CCNP, CCIP
Network Engineer, CableONE Internet Services
"-----Original Message-----
From: Lamar Owen [mailto:lowen@pari.edu]
Interestingly enough, we've tried to do H.323 with some folks on a CMTS con=
nection, and have yet to succeed in smooth video. My testing on my home DS=
L, back when it was 1.5M/.5M (we got two free upgrades; the first one was t=
o 5/.5 and the second to 7/.5) and our main link was an OC3 to a different =
provider, went well. Never really figured out what it was causing the prob=
lems with the CMTS users; the effect was that the H.323 session would start=
up and negotiate at 384Kb/s, and a few seconds of video would traverse fin=
e, and then the link would start dropping more and more frames until it die=
d entirely; my testing on my slower DSL didn't have this problem, and trace=
route showed an equivalent number of hops between. The CMTS connection in =
use was an 8M down 1M up link."
The problem is probably not the connection speed, but congestion on the CMT=
S. If the downstream is saturated (too many people watching Netflix on a no=
de) the available shared bandwidth may not be enough to support your real-t=
ime traffic. Which is a pretty good archetype for the discussion anyhow.
