dslreports logo
site
spacer

spacer
 
   
spc
story category
Examining Comcast's New Bandwidth Management System
Heavy users to be placed on special QoS 'Bus'
by Karl Bode 12:20PM Monday Sep 22 2008 Tipped by Jon See Profile
As we've reported, in addition to a new 250GB cap starting October 1, Comcast has been testing a system that throttles users back to "above DSL speeds." These trials have been conducted for several months in Comcast's Chambersburg, Pennsylvania, Warrenton, Virginia, Colorado Springs, Colorado and East Orange and Lake City Florida markets. While the company has promised transparency in this new management technique, they'd yet to offer specifics as to what customers had to do in order to be throttled.

Click for full size
That's changed somewhat with Comcast's latest filing (pdf) with the FCC, which goes in to significant detail concerning network architecture, and how exactly Comcast plans to target high-consumption users. Previously, Comcast was using Sandvine hardware to forge customer TCP packets, resulting in degraded upstream BitTorrent connectivity for all users, regardless of consumption. You'll note their filing on existing practices (pdf) with the FCC avoids using the phrase "packet forgery."

The new plan is to -- for a change of pace -- actually target the users causing the congestion. According to Comcast, they'll be deploying new hardware and software close to the company's Regional Network Routers (RNRs). This hardware will flip a user from the standard "Priority Best-Effort" traffic (PBE) to lower quality of service (QoS) "Best-Effort" traffic (BE) if a particular CMTS port is congested, and if that user has been identified as a primary reason why. What's Comcast's definition for one of these users? From the filing:
quote:
Following lab tests, simulations, technical trials, customer feedback, vendor evaluations, and a third-party consulting analysis, we have determined that the appropriate starting point for the User Consumption Threshold is 70 percent of a subscriber’s provisioned upstream or downstream bandwidth, and that the appropriate starting point for the User Consumption Duration is 15 minutes.
In other words, if a heavy user on a congested CMTS is using seventy percent or more of their allotted up or downstream bandwidth for more than fifteen minutes, they're tagged as a glutton and treated as a second class customer until his or her consumption eases up. That user has their QoS lowered until their usage drops to 50 percent of their provisioned upstream or downstream bandwidth for "a period of approximately 15 minutes". Note that if you're just a heavy user on an uncongested portion of the network, no action is taken. Also worth noting: upstream and downstream bandwidth is both managed separately.

Depending upon the level of congestion in the CMTS port, this designation (BE) may or may not result in the user’s traffic being delayed or, in extreme cases, dropped before PBE traffic is dropped.
-Comcast
While Comcast has said users will be throttled to "above DSL speeds", they're still not particularly clear on exactly what speeds a throttled user will see. Being tagged a lower class "BE" user "may or may not result in the user’s traffic being delayed or, in extreme cases, dropped before PBE traffic is dropped," according to the filing. Comcast pulls out a bus metaphor to explain the difference between being a throttled "BE" and an unthrottled "PBE" user. Apparently, kids placed on the "special bus" should still be treated fairly:
quote:
If there is no congestion, packets from a user in a BE state should have little trouble getting on the bus when they arrive at the bus stop. If, on the other hand, there is congestion in a particular instance, the bus may become filled by packets in a PBE state before any BE packets can get on. In that situation, the BE packets would have to wait for the next bus that is not filled by PBE packets. In reality, this all takes place in twomillisecond increments, so even if the packets miss 50 “busses,” the delay only will be about one-tenth of a second.
Some of this is still a little murky, and you get the idea Comcast won't have greater specifics on real world speed impact until they've wrapped up throttling trials and get this running on more congested portions of the network. For their part, Comcast is insisting that in trial markets (which you can guess probably weren't Comcast's most congested), less than 1 percent of customers had their traffic managed during a regular day.

What's more, Comcast claims they saw no complaints in trial markets. "To date, Comcast has yet to receive a single customer complaint in any of the trial markets that can be traced to the new congestion management practices, despite having broadly publicized its trials," says the company. Comcast says that should they change the criteria used to determine congestion, they'll notify both the public and the FCC two weeks ahead of time.


199 comments .. click to read

Recommended comments




jlivingood
Premium,VIP
join:2007-10-28
Philadelphia, PA
kudos:2

2 recommendations

reply to FFH

Re: Good Start

said by FFH:

said by Matt3:

said by iansltx_ :

Or watching HD video. An HD download of more than fifteen minutes would pretty much put you over the limit. Or a regular iTunes movie download.
No it wouldn't ... did you look at the flowchart? This only checks your UPSTREAM utilization .... there is no mention of downstream usage.
There was no download flow chart provided by Comcast. BUT, they did say that they would also apply QOS mechanism to download as well. It is even less likely on downloads, however, that prioritiztion would often occur.
We just gave the example of upstream in the chart to demonstrate the concept visually. It is also possible with downstream, with the main difference being that the CMTS Port Utilization Threshold is higher -- it is 80% instead of 70%.

Jason
Comcast
National Engineering & Technical Operations
--
JL
Comcast


jlivingood
Premium,VIP
join:2007-10-28
Philadelphia, PA
kudos:2

2 recommendations

reply to tiger72

Re: It seems to me...

said by tiger72:

makes ya wonder what it would cost them to just upgrade their network instead...
The new congestion management system does NOT substitute for or obviate the need for normal network upgrades - so you are correct. Growing and upgrading our network continues as a normal course of business.

We continue to improve the network in accordance with our plans, including node splits, our rollout of DOCSIS 3.0, etc. One of the reasons we believe our new congestion management practices affect so few users* is because our capacity management model and processes are working well.

* The average percentage of customers that had their traffic managed on any particular day ranged from 0.09% to 0.60% in our trial markets. There were some days in each market where no customer traffic was affected.

Jason
Comcast
National Engineering & Technical Operations
--
JL
Comcast


Matt3
All noise, no signal.
Premium
join:2003-07-20
Jamestown, NC
kudos:12

2 recommendations

Good Start

I think this is a good start. It's going to hit some people who transfer large files pretty hard I bet ... tele-workers, digital photographers ....
--
Linux Haters Unite!