The problem historically isn’t that streaming services are paying for fast lanes but that they have to pay not to be throttled below normal traffic. In other words, they have to pay more to be treated like other traffic.
Even crazier is remember that there are actual peering agreements between folks like cogentco, Level 3, comcast, Hurricane Electric, AT&T, etc. What comcast did that caused the spotlight was to bypass their peering agreement with Level 3 and went direct to their end customer (netflix) and told them they’d specifically throttle them if they didn’t pay a premium which also undermined Level3’s peering agreement with Comcast.
Peering agreements are basically like “I’ll route your traffic, if you route my traffic” and that’s how the Internet works.
Netflix and im sure the other services also have “netflix in a box” media servers that they drop in these peering exchanges and CDN edge datacenters in order to get their media as close to the customers as possible.
The basically bend over backwards to cause ISPs the least amount of traffic, and its still not enough.
I was trying to find the old Level 3 blog post but didn’t because I believe they basically said that Comcast needed to upgrade its infrastructure and never did. Netflix was the cashcow they saw to essentially make them pay for it. As a Comcast customer, I see it as charging the customer twice – first for the Internet service for the content and again because Netflix is going to pass that extra cost onto you (and everyone else who isn’t a Comcast customer).
You’re right on about CDNs and edge / egress/ingress PoPs. It also keeps it cheaper for the likes of Netflix/Amazon/etc. in the long run with the benefits of adding more availability.
I found this wikipedia article about backbones and peering but it really isn’t that great but in the results it also came up with this pretty good presentation from Carnegi Mellon. I was only going to browser a few of the slides but the information isn’t really all that much and the illustrations are good. I think Prof. Nace did an excellent job here. Much better than I would have.
The problem historically isn’t that streaming services are paying for fast lanes but that they have to pay not to be throttled below normal traffic. In other words, they have to pay more to be treated like other traffic.
Even crazier is remember that there are actual peering agreements between folks like cogentco, Level 3, comcast, Hurricane Electric, AT&T, etc. What comcast did that caused the spotlight was to bypass their peering agreement with Level 3 and went direct to their end customer (netflix) and told them they’d specifically throttle them if they didn’t pay a premium which also undermined Level3’s peering agreement with Comcast.
Peering agreements are basically like “I’ll route your traffic, if you route my traffic” and that’s how the Internet works.
Netflix and im sure the other services also have “netflix in a box” media servers that they drop in these peering exchanges and CDN edge datacenters in order to get their media as close to the customers as possible.
The basically bend over backwards to cause ISPs the least amount of traffic, and its still not enough.
I was trying to find the old Level 3 blog post but didn’t because I believe they basically said that Comcast needed to upgrade its infrastructure and never did. Netflix was the cashcow they saw to essentially make them pay for it. As a Comcast customer, I see it as charging the customer twice – first for the Internet service for the content and again because Netflix is going to pass that extra cost onto you (and everyone else who isn’t a Comcast customer).
You’re right on about CDNs and edge / egress/ingress PoPs. It also keeps it cheaper for the likes of Netflix/Amazon/etc. in the long run with the benefits of adding more availability.
Do you have a link to an article or a Wikipedia page that I could read more on this?
I found this wikipedia article about backbones and peering but it really isn’t that great but in the results it also came up with this pretty good presentation from Carnegi Mellon. I was only going to browser a few of the slides but the information isn’t really all that much and the illustrations are good. I think Prof. Nace did an excellent job here. Much better than I would have.