nanog mailing list archives
Re: Some truth about Comcast - WikiLeaks style
From: Adam Rothschild <asr+nanog () latency net>
Date: Wed, 15 Dec 2010 17:47:09 -0500
On 2010-12-15-12:15:47, Kevin Neal <kevin () safelink net> wrote:
Also assuming the backbone and distribution upgrades required between their data centers and their customers costs nothing. It's not free to get bandwidth from Point A (port with TATA) to Point B (Customer).
I don't see how this point, however valid, should factor into the discussion. Missing from this thread is that Comcast's topology and economics for hauling bits between a neutral collocation facility and broadband subscriber are the _same_ whether they ingest traffic by way of a settlement-free peer, customer, or paid transit connection. Speaking to Richard's earlier observations, we too have run into issues attempting to deliver content by way of Comcast's Tata transit, dating back to July of this year. (It's possible the issues might have begun sooner, however this is as far back as our analytics go. I've actually been spending some time documenting how we've been measuring this loss, and how folk might measure it on their production infrastructure utilizing policy routing, routing-instances, and the like -- any interested content folk are welcome to contact me off-list. Suffice it to say, configs are the easy part, the hard part is building a statistically valid sample set without degrading connectivity for paying customers...) Whatever the cause, five months should be ample time to turn up some additional transit capacity or otherwise work around the issues; we're talking commodity transit ports in neutral facilities, such as Equinix sites, after all. What we have here is Comcast holding its users captive, plain and simple. They have established an ecosystem where, to reach them, one must pay to play, otherwise there's a good chance that packets are discarded. Alternate paths simply aren't there, given the no-export communities deployed. As it stands, I could multi-home to NTT, Telia, Tata, and XO, and still get stuck with no good paths to Comcast. While this has happened before (see: DTAG, FT, ...), this is probably the first we've seen it occur in the United States, at scale. Folk in content/hosting should find this all more than a little bit scary. -a
Current thread:
- Re: Some truth about Comcast - WikiLeaks style, (continued)
- Re: Some truth about Comcast - WikiLeaks style Mikael Abrahamsson (Dec 13)
- RE: Some truth about Comcast - WikiLeaks style Rettke, Brian (Dec 13)
- Re: Some truth about Comcast - WikiLeaks style Craig L Uebringer (Dec 14)
- Re: Some truth about Comcast - WikiLeaks style Ricky Beam (Dec 14)
- Re: Some truth about Comcast - WikiLeaks style Laurent GUERBY (Dec 15)
- RE: Some truth about Comcast - WikiLeaks style Randy Epstein (Dec 15)
- RE: Some truth about Comcast - WikiLeaks style Laurent GUERBY (Dec 15)
- Re: Some truth about Comcast - WikiLeaks style ML (Dec 15)
- Re: Some truth about Comcast - WikiLeaks style Jared Mauch (Dec 15)
- Re: Some truth about Comcast - WikiLeaks style Kevin Neal (Dec 15)
- Re: Some truth about Comcast - WikiLeaks style Adam Rothschild (Dec 15)
- Re: Some truth about Comcast - WikiLeaks style Jack Bates (Dec 15)
- RE: Some truth about Comcast - WikiLeaks style Rettke, Brian (Dec 15)
- Re: Some truth about Comcast - WikiLeaks style Richard A Steenbergen (Dec 15)
- Re: Some truth about Comcast - WikiLeaks style Valdis . Kletnieks (Dec 16)
- Re: Some truth about Comcast - WikiLeaks style Jack Bates (Dec 16)
- RE: Some truth about Comcast - WikiLeaks style Rettke, Brian (Dec 13)
- Re: Some truth about Comcast - WikiLeaks style Mikael Abrahamsson (Dec 13)
- Re: Some truth about Comcast - WikiLeaks style Jeff Wheeler (Dec 15)
- Re: Some truth about Comcast - WikiLeaks style Dave Temkin (Dec 16)
- Re: Some truth about Comcast - WikiLeaks style Jeff Wheeler (Dec 16)
- Re: Some truth about Comcast - WikiLeaks style Dave Temkin (Dec 16)
- Re: Some truth about Comcast - WikiLeaks style Owen DeLong (Dec 16)