dslreports logo
site
spacer

spacer
 
   
spc
story category
MLab Data Highlights the Evolution of ISP Throttling
And How Consumer Complaints Changed ISP Practices
by Karl Bode 10:27AM Friday Oct 21 2011
A few years ago Google, in cooperation with academic researchers, the New America Foundation and the PlanetLab Consortium, launched Measurement Lab (M-Lab). Measurement Lab is a suite of tools dedicated to helping broadband customers determine what kind of throttling or protocol discrimination their ISP is employing, presumably letting them make an intelligent decision when shopping between ISPs -- assuming they have a choice. This week the outfit published a new data sat (via Torrent Freak) highlighting consumer test results from between April 2008 and May 2010.

Click for full size
The data highlights how Comcast, who was throttling half of all P2P connections a few years ago, last year only throttled around 3% of P2P connections running on their network. Consumer anger, bad press, Florida's Attorney General and an FCC wrist slap forced the carrier to implement a more intelligent network management system, after their heavy-handed tactics were exposed first here by Broadband Reports users.

Cox, who we noted in 2007 was doing the same packet forgery practices as Comcast yet managed to dodge negative press attention -- also implemented more intelligent solutions in the last few years and saw overall throttling rate decline accordingly -- going from 51% of all P2P connections throttled to 3 percent.

Initially, many ISPs used network throttling as a substitute for network capacity investment. Many of these carriers ultimately replaced systems that throttled all users -- to systems that intelligently targeted only the heaviest users on the most congested nodes. The changes are a testament to the fact that consumer complaints (this 2007 Broadband Reports forum thread started everything) can have a massive impact on ISP policy, even if many ISPs still aren't clear on exactly what kind of a connection a consumer is buying.

As of the data sets last date, the only U.S. ISP left throttling more than 10% of P2P connections on their network is wireless operator Clearwire. Clearwire, whose network management practices recently placed them in last in a ranking of Netflix streaming quality, throttled 17% of P2P connections according to the last data set recorded. The rankings highlight the throttling of all ISPs globally, and users who have watched Rogers heavy-handed tactics over the years (which have gone so far as to include throttling all encrypted traffic on their network just to stop P2P users) won't be surprised to find Rogers on the top of the list.

view:
topics flat nest 

spanglo
Premium
join:2004-05-17
San Diego, CA

Confused

All of the Cox authorized support representatives in the Cox HSI forum have earnestly denied the use of bittorrent throttling over the years. Interesting.

Karl Bode
News Guy
join:2000-03-02
kudos:39

Re: Confused

Well, technically they have used "traffic prioritization and protocol filtering..." so they're partially right. They were initially doing the TCP packet reset shenanigans Comcast was doing early on...

borv
Onemhz On Aim

join:2000-10-06
Astoria, NY

Re: Confused

How exactly were they doing this? Do you have any details of this? I'm curious.

I've been having strange issues with my TWC road runner service and am trying to find out if they are running a similar scam. Please look at a thread I made concerning the issue:

»[TWC] strange disconnect issue twcnyc - larger transfers

dvd536
as Mr. Pink as they come
Premium
join:2001-04-27
Phoenix, AZ
kudos:4
said by spanglo:

All of the Cox authorized support representatives in the Cox HSI forum have earnestly denied the use of bittorrent throttling over the years. Interesting.

I don't know what they're doing but i cant get any eztv torrents to start.
--
Oh YES! let me drop everything i'm doing regardless of who it affects to deal with your petty little problem!
mj3431

join:2003-04-21
STL, MO

Interesting

Interesting to see Comcast and Charter at the top of this list, while Netflix lists them as being top performers for streaming.

FFH5
Premium
join:2002-03-03
Tavistock NJ
kudos:5

Re: Interesting

said by mj3431:

Interesting to see Comcast and Charter at the top of this list, while Netflix lists them as being top performers for streaming.

The list you looked at was for 2008. If you used the tools and looked at 2010 numbers you would have seen that Comcast was among the very best at NOT throttling.


Karl Bode
News Guy
join:2000-03-02
kudos:39

Re: Interesting

Yes, pressing play highlights Charter's quick trip from worst to best...

newview
Ex .. Ex .. Exactly
Premium
join:2001-10-01
Parsonsburg, MD
kudos:1
Reviews:
·DIRECTV
·Comcast
said by FFH5:

... Comcast was among the very best at NOT throttling.

That's because the got their pee-pee whacked hard in the Press and by their subscribers. If that hadn't happened, I'm certain Comcast would have ramped UP their throttling of other protocols.
Trencher

join:2007-02-12
Etobicoke, ON

Rogers

OUCH... Now if only the mainstream press would pick this up and plaster it everywhere. Might get a bunch of people to drop Rogers and force them to change their tune.
nweaver

join:2010-01-13
Napa, CA

=3% is probably measurement error

The Glasnost style testing does have a false positive rate, when noise (eg, such as on a WiFi) causes the BitTorrent traffic to trop but the normal traffic to stay unaffected. The 3% "interference" rate is probably due to measurement error.

(The Glasnost style test uses differential traffic to the servers: some looks like BitTorrent, some like HTTP traffic, which means it shouldn't be affected by Powerboost or other traffic-agnostic shaping.)

Which means we can conclude that pretty much everybody but Clearwire is no longer mucking with P2P traffic.