Live data from Hacker News

Netflix Agrees to Pay Comcast to End Web Traffic Jam

online.wsj.com

311–320 of 351 posts

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#311
post #44

Earlier quoted context omitted.

That is not going to happen. The onus is not on the consumer, it's on the service provider. The service provider may transfer the costs onto the consumer but Comcast won't directly charge the consumers a premium. That would be a bad move and Comcast would lose out its customers to competitors.

The reason Comcast would rather have Netflix pay than pass those costs to consumers is because they don't want to make net neutrality any more salient for consumers. If Netflix pays Comcast then it goes mostly unnoticed by the average person. If people have to start paying then the people screaming for net neutrality will start to scream louder, and Comcast really does not want that. It's not about competition with o…

Will that really hold up, though?

Netflix isn't just going to absorb those payments, they'll have to pass them along to customers. Customers will want some explanation for why their bill is going up . And if Netflix's per-user costs become substantially different at different ISPs, they'll have to start thinking about ISP-specific surcharges...

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#312

Earlier quoted context omitted.

Its silly to pretend that the abstraction is the reality. There is no Internet, just a bunch of private networks connected together. No amount of wishful thinking will change that underlying reality. Now that real money is in play, the fiction of the internet is crumbling and the physical nature is becoming exposed.

"Its silly to pretend that the abstraction is the reality" No more silly than pretending the the abstraction of "files" and "directories" is reality. No more silly than pretending that your computer is running a Python program, when in reality there are sequences of machine language instructions being run. Computers are useful because of abstractions. "There is no Internet, just a bunch of private networks connected…

> No more silly than pretending that your computer is running a Python program

Its a useful abstraction, but as physics causes CPUs to hit a performance wall on single threaded code, then you start banging up against the GIL and the failure of the abstraction to deal with parallelism. And when that happens you can't bury your head in the abstraction of a Python program. I'm not saying that abstractions aren't useful, but at the end of the day you have to concede that they are merely abstractions. When the abstraction becomes disconnected from the reality, reality wins.

> In short, had we told these companies to adapt or die rather than tolerating their efforts to roll back the clock, we would not be in the situation we are in today.

Adapt to what? Abstractions must adapt to reality. Reality does not have to adapt to abstractions. You can of course invoke the government, to regulate the reality into conforming to the abstraction, to the benefit of certain companies and the detriment of others. But history has not been kind to regulatory regimes of that nature. The deep irony of your position is that its the same thinking that led to the current mess with cable. In the 1980s, cable was the future. People invoked the potential to distribute educational material and public access material to the masses, and decided that such an important technology should be regulated. So they created the cable monopolies, under the conditions that they have cheap basic access rates ($13/month in Phila), carry public access or educational programming, and build out even to poor and rural communities. And it was an utter disaster in retrospect.

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#313
post #285

Earlier quoted context omitted.

What if, say, ISP1 and ISP2 both have shitty performance of netflix and then netflix pay for ISP1 to make it better. They do not pay ISP2. So, essentially netflix is available only for ISP1 and not ISP2. That sounds like that the parent comment is saying and what could happen here.

Why would Netflix want to do that?

Because maybe ISP2 wants more money, or Netflix can't afford to pay everyone off.

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#314

Earlier quoted context omitted.

This is exactly the opposite of what you are talking about - there is no preference being provided for Netflix traffic. There is no charge going to the customer for access to Netflix. This is simply the case of a provider, discovering that Cogent provides poor transit services, realizes that you actually have to pay in order to get good Internet peering. This has always been the case on the Internet. As long as Comca…

According to the article, "the average speeds of Netflix's prime-time streams to Comcast subscribers had dropped 27% since October" I guess the question is, did all other traffic see the same drop? And what changes in their network would cause this?

Actually yes, anecdotally. I have a dedicated server in a DC and Comcast at home, and my SCP speeds went down over the last few months to the extent that I had to move DCs to another with no Cogent.

Tata is also spectacularly bad with Comcast.

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#316
post #252

Earlier quoted context omitted.

Peering agreements existed in the form they did because of the underlying assumption that traffic would be roughly symmetric. Services like Netflix that send massive amounts of traffic in a single direction violate this assumption. This is one of the underreported issues in the whole net neutrality debate. In the late '90s I worked in the WAN engineering group doing capacity planning at a big regional CLEC, which act…

DS3 line? What is this, 1996? Each individual Comcast customer has in theory paid for a DS3 worth of bandwidth (modulo the "cap"), which should be a reality-check that some of your assumptions from that era need updating. The defense (and it was always a thin one) of ratio clauses in peering agreements was hot potato routing. The ingress traffic is what you actually have to haul a long distance, so the agreement says…

I used DS3 as an example. But no, even in 2014, you've only paid for a DS3 (or whatever) worth of bandwidth in theory. It's not just the last mile that's oversubscribed -- so are all the trunks leading up to it. If that's changed between then and now, I would lay pretty good money that it's changed to make things worse, not better.

Without meaning to express either approval or cynicism, market power has always been the only thing that matters in this business. But CDNs have paid big ISPs special rates for years, because they really do put special strains on infrastructure. And my impression remains that that's what this particular battle is ultimately about.

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#317
this. is. awesome!

hollywood pressures comcast.

comcast shakedown netflix.

netflix offers money to comcast, comcast and hollywood is pleased.

so far it sucks... but then:

netflix pass that as tax to user.

users flee from comcast.

new providers appears.

hollywood cant shake them down.

for the first time, ISP in america have a truly competitive market.

the end.

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#318
post #132

Earlier quoted context omitted.

Why do you think it would drive down Netflix costs? In old world there were a dozen of "tier1" providers who would all peer with each other more or less settlement free. It is conceivable that one of these providers (I think it was Cogent in this case) was offering " low rates " to take Netflix traffic and deliver that traffic to Comcast over their settlement free peering. As traffic flow between Cogent and Comcast w…

Peering agreements existed in the form they did because of the underlying assumption that traffic would be roughly symmetric. Services like Netflix that send massive amounts of traffic in a single direction violate this assumption. This is one of the underreported issues in the whole net neutrality debate. In the late '90s I worked in the WAN engineering group doing capacity planning at a big regional CLEC, which act…

you are some 5 yrs late... ISPs already put usage in the contracts.

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#319

Earlier quoted context omitted.

This is exactly the opposite of what you are talking about - there is no preference being provided for Netflix traffic. There is no charge going to the customer for access to Netflix. This is simply the case of a provider, discovering that Cogent provides poor transit services, realizes that you actually have to pay in order to get good Internet peering. This has always been the case on the Internet. As long as Comca…

According to the article, "the average speeds of Netflix's prime-time streams to Comcast subscribers had dropped 27% since October" I guess the question is, did all other traffic see the same drop? And what changes in their network would cause this?

Verizon customers have had an ongoing similar issue, also with Cogent peering. As I've stated in comments past, I had to threaten to cancel our business FIOS plan because, with 50% packet loss, the route to one of our colocation facilities was effectively unusable.

Re: Netflix Agrees to Pay Comcast to End Web Traffic Jam

#320

Earlier quoted context omitted.

Remember when there used to be five different cable companies listed in the weekly TV listings? Maybe you never saw that in your area, but when I was growing up in a semi-rural suburb, there were many cable TV competitors. Surely they didn't all require franchise agreements to want to enter the market. Further, assuming franchise agreements exist and were necessary, consumers and municipalities should expect better f…

You're totally ignoring the difference between cable systems now and cable systems back then. Building modern HFC networks capable of supporting broadband requires tremendously more investment than cable TV systems of yore.

You said "wired TV choice".
Post reply on HN