Question on how to look for throttling.

[Callie Del Noire]

I got comcast and use AppleTV/Netflix a bit..

And I've noticed..if I run the SAME HD show through Comcast.. it buffers in 20x faster than the SAME HD video does from Apple and about 10x faster than netflix.

I just got off the phone with a VERY smug technician who told me it was all on my side. First it was the modem. Reset it.. pings really well on my pc.. next. Then it was CLEARLY my splitter/cable. I told them I use a low impedence splitter and TDR'd my cable line. It's good. I even pointed out that my impendence was about 1/10th of the comcast splitter they had originally given me. This was the point where the tech started sounding uncertain. (Clearly the 'how to BS a customer' binder didn't cover someone who troubleshoots wiring for a living in the past or knew what their network db loss and data rate was. Or how much of their alloted bandwidth they had used)

I'm really really good on the wiring.. learning my network tools. What I'm looking for is ways to definitively PROVE that they are throttling.

Anyone know of any tools that can do that?

[Tamhansen]

throttling is very sneaky. If you are speaking about throttling specific sites, this is very hard for ISP's to do, and rather useless as well. It is more likely the problem lies with the bandwidth provided by the uploading site. Netflix being extremely popular will at times be very slow simply due to server load.

If you want to know if certain types of traffic are being shaped or throttled, Glasnost has tests you can try. It will simulate the kind of traffic and give you results you can compare. Gthey can never be 100% sure but it will give you an idea on where to look. http://broadband.mpi-sws.org/transparency/bttest.php

[Callie Del Noire]

You know I would buy it EXCEPT that I have two friends who don't have Comcast who don't get those results. The Iron Man 2 trailer streams almost INSTANTLY from the Apple TV store. In line filtering for specific commercial domains like Netflix, Apple and Hulu wouldn't be that hard on the server side.

Glasnost was one of the tools I checked out but it doesn't look for domain specific filtering.

[Ephiral]

throttling is very sneaky. If you are speaking about throttling specific sites, this is very hard for ISP's to do, and rather useless as well. It is more likely the problem lies with the bandwidth provided by the uploading site. Netflix being extremely popular will at times be very slow simply due to server load.

I'm on unthrottled fiber. I have never had (US) Netflix take longer than 15-20 seconds to buffer, unless the problem was with my LAN. This includes peak times.

[Tamhansen]

hmm. yeah as I said it's tough to look for specific domain throttling, especially as this is most likely done on the side of their servers. I.E. they limit the bandwidth to their servers, then leave the connection between their network hub and your end connection flowing freely. Any tool would then simply show it as being a bandwidth issue with the uploader. I guess the only thing you can do is get a comparison of your traffic with someone who has a comparable package from another provider.

[Callie Del Noire]

Yeah.  I'm SURE it's not my hardware and they can't BS me about my cables.  And I know that if I do a HD EP on their service it starts instantly BUt if I do the same EP on another for pay service it starts to lag. Apple TV and Xbox are the worse BUT it hits Hulu and Netflix a lot.

[Vekseid]

I got comcast and use AppleTV/Netflix a bit..

And I've noticed..if I run the SAME HD show through Comcast.. it buffers in 20x faster than the SAME HD video does from Apple and about 10x faster than netflix.

I just got off the phone with a VERY smug technician who told me it was all on my side. First it was the modem. Reset it.. pings really well on my pc.. next. Then it was CLEARLY my splitter/cable. I told them I use a low impedence splitter and TDR'd my cable line. It's good. I even pointed out that my impendence was about 1/10th of the comcast splitter they had originally given me. This was the point where the tech started sounding uncertain. (Clearly the 'how to BS a customer' binder didn't cover someone who troubleshoots wiring for a living in the past or knew what their network db loss and data rate was. Or how much of their alloted bandwidth they had used)

I'm really really good on the wiring.. learning my network tools. What I'm looking for is ways to definitively PROVE that they are throttling.

Anyone know of any tools that can do that?

They aren't, necessarily, throttling it on your end.

Google for example hosts its content at your ISP directly:
- http://mitchribar.com/2013/02/how-to-stop-youtube-sucking-windows-guide/

The reason why it so frequently is slow is because these servers often have a very limited amount of space for all the traffic they serve, frequently caching things for mere minutes.

173.194.55.0/24 is specifically Google's Anycast range, from the looks of it.
206.111.0.0/16 is XO Communications, which is a connectivity provider (E.g. one of our current host's peers is XO). It's likely that Google is far from their only customer using this range and I read people talking about this speeding up Netflix as well, which does not surprise me.

While Mitch's site blocks both rules, I'd really only use one at a time and see how things work. My guess is XO is a lot less likely to know wtf it's doing about serving content than Google. They're a high speed, budget connectivity provider. So my recommendation would be to test -only- 206.111.0.0/16

[Vekseid]

In honor of who is actually more likely to blame for this, I decided to name the rule a bit differently:

Code Select Expand


netsh advfirewall firewall add rule name="XOCOMMSUCKS" dir=in action=block remoteip=206.111.0.0/16 enable=yes


Ahem.