[Cerowrt-devel] Network tests as discussed in Washington, DC
dpreed at reed.com
dpreed at reed.com
Sat Nov 10 09:18:07 EST 2012
I follow this pretty closely. The person you want to talk to is Sascha Meinrath, about M-Lab - and if you have a good proposal, he has money to fund a measurement project.
M-Lab has been a partial failure politically. The reason is this: the FCC does not want to do anything that would annoy incumbents running current Internet access systems. Direct measurements of latency are too dangerous. If you speak to the only Commissioner who ever tried to get unbiased measurements (Copps) you will find that his attempts were sabotaged on every side.
The same thing happened in the CRTC inquiry regarding Net Neutrality, where I testified as an "expert witness" on the argument that prioritization was necessary to reduce congestion. My testimony pointed out that congestion was best measured by direct latency measurements. The incumbent operators *refused* to provide latency data, claiming a) they never measured it, and b) latency can be calculated from average traffic volume measurements by "Little's Lemma", so it was unnecessary. (this last is categorically false, but it was "one expert vs. another").
I wish you godspeed in getting direct latency measurements publicized. The dataset collected by Netalyzr has been attacked by the incumbent operators pretty strongly as "biased". (of course the measurements the FCC commissioned to a contractor recommended by a Congressional staffer are "unbiased").
From: "Dave Taht" <dave.taht at gmail.com>
Sent: Saturday, November 10, 2012 5:54am
To: "bloat" <bloat at lists.bufferbloat.net>, cerowrt-devel at lists.bufferbloat.net, "bloat-devel" <bloat-devel at lists.bufferbloat.net>
Subject: [Cerowrt-devel] Network tests as discussed in Washington, DC
The RRUL test idea presented earlier this week would do best with many
underlying servers, instrumented with TCP stats collection, and a
large backend for analysis, and thus I got interested in the current
state of affairs in the internet as to how to pull that together in
conjunction with larger labs and universities.
So, I started sorting through the debates at the FCC about network
testing. A wide range of opinion is presented, AND represented - I
randomly clicked on the names I recognised (like isc, verizon, karl
auerbach, measurement labs, new horizon foundation, and many, many
others. I'm pretty sure to get a balanced view I need to click on the
names I don't recognise!)
see position papers here:
After reading a dozen, I sighed, and went back to coding.
I'm just this guy, trying to fix bufferbloat, by all technical means
necessary. Additionally we're working on fixing ipv6, home routers,
home gateways, naming, etc - and I rarely pay attention to politics!
If anyone can provide a summary here of the debate as it stands now
and who the players are, it would be enlightening.
In other news:
I was very pleased to hear that Srikanth Sundaresan of the Bismark
project ( http://projectbismark.net/ ) won the "Applied Networking
Research Prize" at this past week's ietf ( http://irtf.org/anrp ),
"Broadband Internet Performance: A View From the Gateway"
An early version of cerowrt was used in some versions of bismark (they
later went to openwrt stable). I continue to build their repos for
possible use in cerowrt (or vice versa), and I'm contemplating
extending their open sourced backend database schema to include data
from the rrul tests....
Fixing bufferbloat with cerowrt: http://www.teklibre.com/cerowrt/subscribe.html
Cerowrt-devel mailing list
Cerowrt-devel at lists.bufferbloat.net
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Bloat-devel