Nice costumers you have there SK Broadband, would be a shame if your costumers leave you. Oh? They are unlikey with Netflix not in a monthly cancelable contract but Internet plans generally require contracts of 2 to 3 years in South Korea.
Nice chaebol you have there, would be a shame if its behavior became the subject of international trade sanctions from the country that is providing you protection against your crazy and heavily armed cousin who lives next door.
I can't speak to South Korea's ISP market, but in the US the major cable providers are now doing very well.
Comcast is at $18b in operating profit, with roughly 18% operating income margins. Charter Communications is at $9.6b in operating income, with 19% operating income margins. Those are the two biggest cable companies in the US.
In the early decades of cable, profitability was notoriously flaky for the industry. Enormous amounts of required capital investment, and it took a long time to saturate - scale into - the US market. Now the market is settled and they're printing consistently giant profits.
Interestingly Charter has become a giant, worth $133 billion, and Paul Allen used to own a controlling interest in the early incarnation of it. It was part of his so called wired world collection of investments.
> How big are the operating margins for major ISPs nowadays?
In 2020, KT: 1.18 trillion KRW (1.00bn USD), SKT: 1.34 trillion KRW (1.13bn USD), LG U+: 886 billion KRW (749m USD). Their revenues range from 10 to 25 trillion KRW for the reference.
Hmm, I guess ISPs could just drop traffic if they can’t afford it, but maybe if you push a lot of data you could pay more to have the content actually delivered to end users.
But that would raise subscription prices. Should every user really pay extra for each subscription based on the data they might use on that subscription?
Akshually ... that could make a bit of sense. Instead of paying for bandwidth you rarely use to your ISP, you could pay for bandwidth you actually use to the content provider.
However, that would leave free content inaccessible, and probably create another wealth divide.
But bandwidth is really cheap and this problem doesn't exist anywhere else. Worst case the ISP can increase their rates by a couple of dollars and improve their cables.
No argument from me. I just saw a tiny bit of logic behind another model of balancing the costs. The ISP would still have to upgrade the infrastructure in the end (although my model could lead to e.g. Netflix buying up ISPs).
That's leads to a pretty simple conclusion, because Netflix doesn't push any data. I pull data from Netflix. I have already paid my ISP to pull data from anywhere I ask. If my ISP sits down at a negotiating table with Netflix and threatens to disrupt that, they should be laughed out of the room. "Pay me, or else I'm going to break my obligations to an unrelated third party." isn't a move that should have any respect.
Then Musk and Bezos should be laughing since they'll likely be getting more customers for Starlink and (eventually) Kuiper. Hopefully low earth orbit satellite internet will put an end to ISP monopolies in the US since the government wont.
The ISP could drop or throttle the traffic, assuming that's allowable under the ToS with their customers. But then of course their customers will quit and find an ISP that doesn't throttle/drop.
Regardless, Netflix is not pushing content. The consumer is pulling it.
As a one-off, issues like the Squid-Game aren't so bad, but if it's a chronic problem for the ISP then it's because their model of consumer consumption was wrong.
I’d be ok with then charging a significantly lower for a “limited bandwidth” connection. But I bought an unlimited, so they have no right to complain.
Imagine if the same principles were used in other areas. You iPhone suddenly locking up after using Facebook to much because Facebook wasn’t paying Apple for access to its users.
"Push" and "pull" in this context refer to the initiation of a session. The user is pulling, because they start streaming. Netflix does absolutely nothing by itself - it doesn't randomly start delivering data to you.
> Netflix said it will review SK Broadband's claim, and seek dialogue and explore ways in the meantime to work with SK Broadband to ensure customers are not affected.
In other words, they will ship some Open Connect Appliance boxes to the ISP.
Some think it’s unfair that big players like Netflix can do so. But I think them doing that is far far better than the idea of ISPs charging content providers for the bandwidth.
Like other said, the customers of the ISPs are already paying for the bandwidth. Having the ISPs charging Netflix for the bandwidth too, that’s no bueno.
This is basically how the internet works though, isn't it? Your network and my network have parties interested in each others data. If parts of your network are generating traffic that I don't think is reasonable then I can either disable their access or negotiate with you or one of the parties on your network to pay more or I can deny or throttle access to the parties on my network. And a OCA box just makes everything better for everyone where it's possible to employ them.
"If parts of your network are generating traffic that I don't think is reasonable"
But it's your customers (the end users) generating traffic by watching content on my network (Netflix shows). If your customers can generate more traffic than you can handle, regardless of its a single entity or multiple entities, then that's a contract issue between you and your customers.
Exactly this. To say that Netflix is generating traffic pointed at SK broadband would not be fair. Those TCP connections are initiated by the end user. Without the user requests there would be no traffic.
So the philosophical question really becomes, what exactly is the end user buying from their ISP?
Wouldn't that also include the bandwidth? Meaning, if the customer is already paying for the bandwidth they will consume, why wouldn't that cover bandwidth usage of Netflix content?
No. That's not what my contract with my ISP says I'm buying.
Besides, physical access alone is useless. Unless you're claiming that ISP's have been selling physical access but giving away usage for free? I don't know if that's the intent of your comment, but again it isn't how my ISP words the contract.
>Unless you're claiming that ISP's have been selling physical access but giving away usage for free?
I don't think zakki is too far off from the truth. Looking at Spectrum's Residential Internet Services Agreement, physical access might be what it boils down to. It's a little shocking they can have terms that make no guarantee to any amount of bandwidth at all, even on "their network". I suspect other large ISPs in the US use the same terminology.
>Subscriber understands and agrees that Spectrum does not guarantee that any particular amount of bandwidth on the Spectrum network or that any speed or throughput of Subscriber's connection to the Spectrum network will be available to Subscriber.
But I think you're correct. Having a physical connection does you no good if it isn't passing packets.
As a consumer I paid my ISP to carry say 50 Gb/s of data. They took my money on the understanding it's their job to carry that data for me. If I only wanted to browse text I wouldn't be paying for 50Gb/s. What did they think I was going to do?
That is a failure of Xbox in my opinion. Games are 100gb these days. Massive. My monthly bandwidth is line 400gb. That is pretty common in Canada. We can’t afford to download all the games, again. There should be a local transfer of games done over wifi. It should be made easy.
Not quite. They have a soft 15 Gb cap, past which they throttle you down significantly but also don't charge you. And then it's $10 / Gb if you want to extend that cap.
There's more than one plan. The one I'm on is $10/GB until you use 6 GB (or $60). You're right though that they slow down after 15 GB (not Gb) but I've never come anywhere near that.
The way I look at it is that my plan is $20 then my data is $1 / 100 MB.
On a mobile connection you generally have a data cap, and you can switch to a more expensive plan for a higher cap. That's the same thing as paying for bandwidth, just that you are buying in bulk with pre-commitment.
To get significant damage as a single individual, take someone on a traditional satellite connection who can't get some important work done on time because ads put him above his data cap.
That is indeed an interesting point, especially for users who have metered Internet service.
Consider a user with a data cap of 1000GB, is metered as consuming say 1200GB of data, of which 200GB is Ads. The user is forced to pay extra for the 200GB to the ISP which the user should not have had to pay if there were no Ads.
Ignoring that ISPs should not meter the service, the above just doesn't seem right. The individual Internet services may be subsidized by Ads, but the user is paying to the ISP. So, in effect a user is paying, just to a different party.
Some lawyer should definitely look into it and see if there is potential for a class-action against Ad vendors.
Edit: I'm sure the flip argument would be that the user requested the Internet services that came with the Ad, so in effect the user is responsible for consuming the data. But that is where I think it becomes interesting and given that IMNAL I don't know what the legal minds think.
My question is why Open Connect Appliance wasn't even in their network in the first place?
I mean there are only a few ISP in every country. SK Broadband isn't small either it is something like 25% of market share. If I remember correctly Korean Telecom was 50% and LG was something like 15%. SK Telecom is also the largest MNO is South Korea. And if you are into hardware you may have at least heard of SK Hynix, world's third largest NAND manufacture.
Normally being a Korean conglomerate or Chaebol I would have thought SK must have some subsidiary competing with Netflix. But this doesn't seems to be the case here.
I wouldn't be surprised if the normal bandwidth is able to be handled by existing connections, rather than the boxes with the ISP's. Is netflix super popular in Korea?
Netflix made a few big splash with content made specifically for South Korean. Or K-Drama. But I dont have any data to judge whether they are super popular or not.
> The popularity of the hit series "Squid Game" and other offerings have underscored Netflix's status as the country's second-largest data traffic generator after Google's YouTube
That too is true. As of March 2021, Netflix was the only over-the-top service that has reached 10M subscribers in SK (seconded by Wavve, currently operated by SKT and but only had ~4M subscribers at that time).
> My question is why Open Connect Appliance wasn't even in their network in the first place?
By having Open Connect Appliance network gives up ability to traffic engineer Netflix, which may decrease the leverage the network has with its peers or its transit providers.
Fundamentally, unless someone must peer with you (settlement or settlement free) and must peer with you on your terms ( locations, types of connections, etc ) you always want traffic that you and only you can control.
No matter what, SK Telecom is going to be eyeball (ingress) heavy and no amount of engineering of Netflix traffic is going to help with that, as it's all inbound anyway. Not that Netflix's upstreams like Level3 or Telia would be peering with SK Telecom anyway. Reducing the amount of Netflix traffic can only help.
Traffic engineering does not just cover what. It also covers where. At the eyeball network level where is way more important than what as what is constant.
> Normally being a Korean conglomerate or Chaebol I would have thought SK must have some subsidiary competing with Netflix. But this doesn't seems to be the case here.
Your guess is indeed correct! Two out of three major ISPs in South Korea operate their own OTT services (Wavve and Seezn). LG U+ is an exception and also the first major ISP to join the Open Connect Appliance program for the obvious reason.
> South Korean lawmakers have spoken out against content providers who do not pay for network usage despite generating explosive traffic
It's impressive how this framing has gotten so much acceptance. The content providers are not generating the traffic, the users are, and they are paying their ISPs to be able to do so.
If anything, customers should be suing ISPs every time they don't get the expected bandwidth. It's the ISP that fails to deliver according to the contract.
You're certainly right, and the specific term here for what consumers should, but generally don't, have is a "Service Level Agreement" (SLA). This is pretty standard for business links and nearly anything at all at the datacenter level, if you ever subscribe to a whole line at a colo you'll probably see that. SLAs tend to spell out not just maximum sustained bandwidth, but 90/95/97/99%, uptime, what if any burst bandwidth is available, any latency guarantees, etc. All of these are different measures that cost different amounts to hit at any given level, and SLAs spell out exactly what is being paid for in numbers that can be monitored fairly easily.
If I could make one simple change to the home broadband market, it would be insisting on a standardized big number consumer level SLA. I certainly don't expect residential rates to cover the same 99.97% or worst case minimum that business class does, that's a major part of what higher end connections are paying for. But whatever they do offer should be transparent, fixed for contract period and measurable. ISPs should by law be required to spell out not just the top line bandwidth level, but also 75%/95% guarantees over a standard time (day/month maybe) and what the worst case guaranteed minimum is. This wouldn't be some big centralized bureaucratic decision about implementations or what numbers have to be hit, but rather merely make sure the market has clear information so customers can compare and ISPs could be held to their promises. Energy Star consumer information might be some guide to layout, but I think something like that would be low hanging fruit.
Also to be clear, a distinction can and should be made for intra-network vs inter-network, and both should be accounted for. Obviously an ISP has the most control and is the most responsible for how traffic traverses their own network to their edges, but ISPs do have some responsibility (affected by size/location) for their peering agreements too. If they're purposefully choking that it's on them, though small/remote ISPs may be at a disadvantage in negotiations there which should ideally have some eyes on it.
I absolutely agree, however national or even regional level transparency reports are bullshit. It needs to be at the exchange / node level. The same ISP will have different levels of service between cities and suburbs.
I agree, but there still need to be constraints. For instance, if I have a 1Gb symmetrical connection, it's probably still unreasonable for me to expect 1Gb connecting to my buddy's 1Gb 3k miles away.
Right, that's likely doable on the same ISP, even 3k miles away.
It does highlight a philosophical question I raised in another comment. What exactly are consumers purchasing from last mile ISPs? Is it 1Gb to anywhere on the providers network? 1Gb to the nearest CDN/Internet exchange?
I don't think there is a good answer for this. But it's important to think about when we talk about ISPs reaching their hand towards large content providers.
It really is wild. ISPs have been overselling "unlimited" connections for forever, and being paid for it, but when their customers actually want to use literally the service they paid for, _legally_, it's somehow wrong?
I guess that's the thing with Netflix though, it's torrent-like traffic, but legal, so there's no excuse to stop it and they have to come up with something. "oh they're generating explosive traffic that's not good"
You seem to spend too much time on leftist web sites.
First, they aren't being paid for unlimited. Do you think that $39.99 or $199.99 per month can cover the cost of transferring unlimited amount of data to another continent?
Of course, it doesn't. Somebody pays $39.99 to download 10 emails, another customer runs BitTorrent 24 by 7 consuming TB's of bandwidth for the same price.
Second, they don't want to sell "unlimited" - they could make more money by selling differentiated services.
The thing that's most ridiculous about this is that the bandwidth is essentially "free" for SK telecom. They sit a netflix caching server in their network and the traffic never leaves. Other than the cost of switch ports (which are dirt cheap relatively speaking) they have absolutely 0 overhead beyond power and cooling for the netflix box which again with their user count is a rounding error.
I'm not sure if the lawmakers are ignorant, bought and paid for, or a little of both but it would be refreshing if one of them took 30 seconds and applied a little critical thinking to the situation.
> The thing that's most ridiculous about this is that the bandwidth is essentially "free" for SK telecom. They sit a netflix caching server in their network and the traffic never leaves.
Wouldn't that require them to MITM the Netflix traffic, which I would hope is under TLS?
Netflix provides the ISP with the box, afaik the box contains a signed certificate for a Netflix subdomain specific to that box. The "real" Netflix servers direct the customer to that subdomain for the actual content.
That server would, of course, have a connection to the Internet, they push logs to AWS and get updates from Netflix's authoritative servers in Netflix's other datacenters, but they also serve HTTPS connections to thousands of ISP users; the upstream traffic would be minimal.
My guess Netflix can guide that by handing back different IPs when the end user requests the netflix server name. They can also handle it in application the netflix application can look and see it is in a network that has a box and just goto that ip/host/proxy instead of the generic one. Netflix still controls the keys. No MITM needed.
Does this mean that Netflix has to have a private TLS key in the box? I wonder how the security of these keys is usually maintained. Interesting engineering/cryptography problem.
This problem might also be an interesting use case for Signed HTTP Exchanges. Sign the video files, then push them to the caching server. The caching server then never possesses the private keys for the connection and also cannot modify the content.
Even then, it's probably too high of latency. These Netflix caching boxes are perennially the subject of "this is how you TLS at line rate and memory bandwidth limits" talks. When you combine this with the fact that these boxes aren't just installed anywhere, but deep inside ISP networks, and ostenisbly the certs aren't for *.netflix.com, but instead for either that specific ISP or even that specifc box, _and_ the certs need to be remotely updated anyway, so they can't just live in an HSM the whole time, I'd just store them on disk if I were Netflix.
And you need the private key for every new TLS connection. And when you are running multiple 100Gb interfaces, that's a lot of TLS connections.
Versus, what's the attack? The ISP with physical access to the box can pull can screw with the cached video data somehow? Or MitM and know what Netflix videos their users are watching?
it's SK, so there's a probably an element that the law-makers are paid for. SK is notorious for sending its past presidents to jail for corruption cases. but yeah with that netflix redbox, traffic shouldn't be taxin' the telco's pockets hard.
Is that actually because SK is corrupt, or because SK is really keen on rectitude in politicians? I think there are more than a few UK politicians who could definitely serve time for corruption, and it's even been occasionally (recently the health secretary) bourne out in court, but it's rarely even a political issue, let alone a criminal one.
Also, the less said about SK's neighbors (Japan, for instance) the better.
There would be a caching server inside SK Telecom's network if SK didn't demand payment from Netflix to provide SK the caching server.
"But SK Broadband demanded that Netflix pay network usage fees on top of installing those caches. The streaming service sued the telco, and that case is pending — Netflix told us that it would continue negotiations even as it made its case in court."
Alternatively, they would peer with SK Telecom but "SK was demanding large content providers money for peering with them."
There is nothing illegal about torrent traffic any more than any other form of traffic. "It's streaming-like, but legal." "It's VPN-like, but legal." "It's cyptolocker-like, but legal." "It's SSL-like, but legal." "It's Tor-like, but legal."
It’s unreasonable to the bottom line of the ISP because it’s high demand on an oversubscribed network. Torrents used to be a convenient scapegoat because they were more than likely carrying content that was infringing copyright.
I am surprised that ISPs haven’t started cracking down on WfH, and start saying it violates their TOC because this is a business use. Public uproar would be immense.
"I am surprised that ISPs haven’t started cracking down on WfH", "Public uproar would be immense."
You listed the reason why yourself, you shouldn't be surprised. Maybe -after- the pandemic, when WfH people can be painted as "tech elite" or something, but not while everyone wants to WfH and it's a mixed bag who can.
Judges implement the law though, so if the contract says "no business use" and this was made clear at the time of purchase -- eg by having business level accounts -- then what can the judge do? It's for the legislators to legislate, not the judges (under English law (UK), common law and equity aside).
Actual, I might be wrong here with how contact law is handled ... anyone?
A related consideration: when we moved into our terraced house the land was under leasehold and the lease specified 'no business use'. We bought the freehold some time back, but I've often wondered what my position might otherwise have been. Presumably breech of the lease conditions would be a tort against the landlord and so I'd have been fine as long as they didn't sue.
I meant specifically in the context of a pandemic when working from home is state mandated in many parts of the world. Arguably even constitutions have been bent.
In the current climate, it would only invite government scrutiny and regulation. Which, once it is in place, is unlikely to go away even if COVID does.
> It’s unreasonable to the bottom line of the ISP because it’s high demand on an oversubscribed network.
No, it's not unreasonable. They sell consumers unlimited high bandwidth connections and at no point are they told they can't actually use the service they're paying for. What's unreasonable is cutting off paying consumers and suing streaming services
when the company's dumb business model starts failing because of their own invalid assumptions. If oversubscribing is such a problem, they should stop doing it instead of blaming consumers for overusing their low capacity network.
How is Netflix traffic torrent-like? It's large volumes of video data received over HTTPS connections. Users perform negligible upload apart from ACK packets.
He meant the volume is arguably similar to torrent leeching, although that comparison is a bit outdated; Today it's easy to get high quality torrents that are orders of magnitude more network-intense than Netflix streaming, while in turn today's Netflix is orders of magnitude more intense that old torrenting
Yeah, these companies sell unlimited connections despite not having the capacity to deliver and then have the gall to be offended when we start using what we pay for. Based on their marketing I should be entirely justified if I wanted to saturate their links at maximum speed 24/7. They assume most people won't do it but they deserve absolutely no sympathy when their assumptions are proven wrong.
Lashing out at Netflix and bringing them to court for their own failings is such an absurd course of action. They should try upgrading their networks instead so that their service stops sucking.
It's even more impressive how the average HN reader can't even comprehend this comment (or can, but somehow manages to insist on their wrong, socialist views) and continues to argue in favor of "network neutrality", even though no two users or service providers are the same.
If Netflix should pay $10 mil a month to Korea's Internet providers, why not HN? If HN should pay less than Netflix, why should the user who only visits HN pay the same Internet access fee as the one who constantly watches Netflix?
This was one of the first wins for IPv6. I had a Hurricane Electric tunnel on my FiOS link and was able to bypass the bottleneck that Verizon set up for Netflix.
Now Netflix blocks Hurricane Electric endpoints and FiOS still doesn't support IPv6. We're nearing the 10 year anniversary of that "coming soon" FAQ entry on their webpage.
This seems like a really popular technique in politics. Frame the problem in a way that makes it seem unjust regardless what the wider implications are.
Online ads and data vacuuming is often presented in a similar fashion. Your browser goes to the advertiser and requests the ad. The advertiser asks for your data. Your browser sends the data to the advertiser. And the way we frame this is that advertisers are stealing our data.
Once a system is opaque and automated enough it doesn't seem to matter how it really works. Netflix is a big entity, therefore we blame Netflix. If Netflix didn't exist then the traffic surge wouldn't exist, therefore Netflix should pay.
It's quite common datacenter prices to incl. bandwidth and data.
Netflix/youtube cases it's the content providers that server the traffic, so most of it is generated by them. The client requests are tiny in comparison.
But that's not the case in the article "Netflix began using =SK's dedicated= line starting 2018 to deliver increasingly larger amounts of data-heavy, high-definition video content to viewers in Korea from servers in Japan and Hong Kong".
Admittedly the article is light on the details, yet I'd presume the line has been created exclusively for Netflix, there is no info if there was an agreement between the parties. There is no reference to the Seoul District Court decision, either.
"Internet"/data between major tube providers (ISP) is very different than paying cable company X for end user access. Some just have agreements to carry each other traffics and some pay to carry their traffic[0]:
Lets compare it with food, if you (client) would not eat so much the producer (server) would not produce as much, no server produces traffic without a consumer.
A decade ago I was with an ISP who took a stand for "Google should pay us for the priviledge of having our users watch YouTube". I believe I wasn't the only one who decided that's the last straw and moved to an ISP that wasn't rejecting peering based on greed.
Thankfully that was a market with good competition. They might have hoped to capitalise on some of that anti-American, anti-BigTech sentiment. But everyone I knew who used them was pissed at these shenanigans. Hilariously enough, I switched to a former governmental monopoly.
Netflix has no association with this ISP, you do. It's your choice to request Netflix streams for delivery through the ISP. Blaming Netflix for this would be like an apartment building blaming Amazon for delivering too many boxes to the mail room.
Don’t miss the past paragraph in the article, “In the United States, Netflix has been paying a fee to broadband provider Comcast Corp (CMCSA.O) for over seven years for faster streaming speeds. https://reut.rs/2Y8wOzb”
Two points:
1. In 2014 Netflix made similar deals with AT&T and Verizon.
2. It is not extortion by the ISP, "pay up or else". But Netflix was simply building up its own CDN-capabilities. They ditched Akamai and Limelight and for that they needed their own server endpoints into the networks.
The framing gets a pass because the majority of "explosive traffic" comes from foreign companies. Google and Netflix have no voice in the Korean lawmaking process. When they go to court, they fall right into the trap of the framing that local companies have already set up.
South Korea is well known for having fast internet, but the internet is only fast within the small country. International connectivity has always been crappy, and getting crappier every year as telcos focus on legal avenues instead of actually laying enough undersea cables to meet the demand. In the past, most Koreans only visited Korean sites because of the language barrier; that's no longer true. But it's going to take a lot of time and money to remedy years of neglect in international connectivity. Especially SK Broadband, who IMO has been the most complacent in this department among the Big 3 telcos, precisely because of their huge domestic market share. Where is the money going to come from? Obviously not by cutting executive bonuses. SK Broadband seems to think that they can extort it out of Google and Netflix who, despite their global prominence, are relatively defenseless in Korean courts.
It's not about what is fair, it's about maximizing profit. And if a company can earn money from both consumers and producers, why not? This is what unrestricted capitalism is like.
IIRC there were some companies that got a slap on the wrist (it was something like Uber or a food delivery company) for charging both consumers and producers a commission, which is illegal (I don't remember the legalese term).
What if ISPs created their own CDNs or someone creates a CDN network behind major ISP providers -- Netflix and YouTube top consumed content seems to be a very good contender for this type of a offering.
There's a big difference in usage patterns: Netflix has a small number of videos, each with huge viewer count. That's eminently cachable. Youtube otoh is the long tail, orders of magnitude more videos with small view counts. Some of it may be cachable, but that's going to shave off on the order of 20% of the traffic cost, whereas in Netflix's case the gain could easily be 99%.
Seems like a good opportunity for Netflix to raise their prices in S.Korea and use a portion of that increase to pay what the courts seem like they're going to demand.
The rest of the increase would go to offsetting the additional burden on Netflix that this will cause by forcing them to treat S.Korea differently than everyone else.
I'm honestly surprised that any corporations are paying that there, but if the government is going to requirement, there's not a lot you can do about it. It's a very bad precedent and I don't see it actually going well for S. Korea. Putting up extra barriers is a way to keep companies out. We usually call it a "tariff", but that's generally for physical goods. And the goal is often to prevent as many of those goods from being imported.
What would S.Korea's internet look like if even 1 of the major players decided to just block the country instead of paying the fees?
> I'm honestly surprised that any corporations are paying that there,
It benefits Netflix in the long term as it raises barriers to entry for disruptive competitors.
If, in order to reach an ISP's customers, you have to pay the ISP for the privilege, then creating new startups is harder and the incumbents can just price it in.
I wouldn't be at all surprised if this became more common elsewhere, as everyone involved benefits. Except us, the customers, of course, but we don't get a seat at this table and the people who should be fighting for us are taking bribes (sorry "campaign donations").
While I see the logic in what you're saying, the Netflix business moat is already much deeper due to A) billion dollar content deals with major providers and B) billion dollar investments into first party content.
Content is king and competitors are already priced out long before minor squabbles over bandwidth.
The disruptor may have an answer for the content problem that Netflix never thought of. It's a competition for attention, after all, and if someone else can provide a more compelling alternative than video for people's attention then that will mess with Netflix' business despite its content deals.
Squid Game is one of those shows you can never "unwatch". It's quite harrowing but strangely addictive and a great critique of modern capitalism. Strongly recommended if you have the stomach for it
I'm dubious of this common posture of "horrible violence in entertainment" is ok because it's a critique of XYZ.
It's like "free guy" (spoiler boring) tries to be a critique of video Game violence by positing the existence of another gaming reality where people are nice to each other. But this other reality is just a Macguffin. We come to watch the violence and mayhem and we come to watch squid game for the same reason.
It is better to own the fact that watching torture porn gives you the jollies rather than posturing that it gives you membership of some anti capalist intelligensia.
It's just Romans and Christians and Lions. The crowd loves it because people love blood.
This violence is unrealistic to the point that it's not disturbing (to me, at least). Someone pulls an undamaged bullet from a guy's forehead with tweezers. Blood is the color of cranberry juice. It's so over the top it's lost all impact.
It's also very simplistic. In subject matter, in character archetypes, sometimes in acting, and most annoyingly, in editing (how it never trusts the viewer to remember what happened 25 minutes ago, so they show flashbacks to remind you).
> It's also very simplistic. In subject matter, in character archetypes, sometimes in acting, and most annoyingly, in editing (how it never trusts the viewer to remember what happened 25 minutes ago, so they show flashbacks to remind you).
I just saw the final episode last night, and I don't remember seeing many flashbacks other than at the start of E01.
I admit it is slower moving than I'd like, and some of the episodes had way to much filler and not enough plot arcs, but I seriously do not remember seeing flashbacks.
It's full of blatant inconsistencies and unrealistic to the point of being comic (and yes I am capable of attenuating my definition of "reality" for shows and movies.) I rarely watch TV but someone made me watch this, and I'm enjoying it as a comedy. Am I in good company or is everyone else taking it seriously?
If I were the judge, I would ask Netflix to stream low bandwidth version of the content so it won't impact SK Broadband.
AND, tell BOTH parties that they must put announcement in their respective website, "SK Broadband provides you internet at a cost (as per their claim). To enjoy Full HD of this content, pay extra to SK Broadband".
Considering how often I've seen the "SK has the best/fastest consumer internet connections" news bits, I'd be awfully angry at the provider for drumming up the hype and failing to deliver.
There are three major nationwide broadband services in SK: Korea Telecom (KT), LG Uplus and SK Broadband (SKBB, a subsidary of SK Telecom). Among them SKBB is considered the worst because both KT and LG U+ operate their own infrastructures due to their pre-privatization history but SKBB don't, so SKBB has to perpetually lease infra from them. It is therefore no surprise that SKBB (and also SKT) is one of the biggest forces against net neutrality in SK, given their service is especially weak and net neutrality will make it worse.
By this logic the consumer would have to pay the ISP fee to every new service they sign up for, like Netflix, Youtube, Disney etc. I prefer the way it works now where I only have to pay for my connection once.
Similarly, Australia had problems when Netflix was first becoming popular here, putting huge strains on the network and causing widespread congestion issues from 6 to 10 pm.
I wonder if we'll ever see a predictive home-caching system; if I watch the first three episodes of squid game in one night, it might be sensible to have my TV download the rest of the series overnight.
We do now. The problem was when Netflix first opened here and the vast majority of people were still on ADSL links or even slower, and our NBN infrastructure was still just starting to come together (which I'm not going to go into details about because omg there's a huge political fight in that which I really don't want to think about)
Short version: our Netflix is just fine now. There were just a couple rocky weeks/months at the start. :)
having multicast working on the internet would make this extremly efficient for the distributors (i.e. netflix) but the last mile (i.e. your ISP) still would have to deliver the packets to their customers one by one. at Netflix scale it would probably be already very beneficial to do that by broadcasting the stream starting every 5 seconds or something for uncached viewing. (i.e make the player wait for the next broadcast and jump to different ones [or start a new one at desired offset if none exist] when seeking).
Instead of using the internet its also feasible to use the DVB network like this. they could even broadcast a copy long before release so its guaranteed to be ready for offline watching when released. i am almost certain Sky Germany does something similar...
a long, long time ago, we used devices called "vcr" to record specific streams that were broadcast over the cable network to have them ready for offline watching.
A hybrid approach would be the best. You could start with an unicast stream and prefetch the rest of the show in the background by subscribing to a few cycling multicast streams.
That requires a few gigabytes of temporary storage of course.
At least the Android app does have the ability to do downloads, and I thought it had the ability to schedule downloads for off hours, but I don't have it in front of me so I might be misremembering
This seems to be the opposite of Internet behemoths zero rating their bandwidth to undercut smaller competitors: Internet behemoths being shaken down for transit technically already paid for. I suppose the ISPs didn't expect this level of utilization.
I think it becomes easier to think about this, if I frame the problem in terms of more tangible objects than bandwidth and data.
The public street network is in large parts paid for by taxes, so for everyone access to the network is already payed for.
If I now envision a single company blocking all streets with its giant truck fleet, I could completely understand a motion to introduce a new form of toll to be payed by such companies to avoid having to rise taxes.
That toll would most likely just increase the price of whatever said company is selling.
I don't think that's a fair framing. As others have said, the users of the broadband pay for their bandwidth. It's not Netflix pushing data, it's the users pulling it. If you have 100 users that pay you for 1Gbps/s but your network can't handle 100Gbps/s that's your failure to adhere to the contract you made with your users.
In the road example, it's more like if every person paid for the ability to put one truck on the road and everyone delegated their truck space to a shipping company who now has millions of trucks hogging the roads. Yes, they're using tons of infrastructure but their usage has been paid for already.
Its more like, there is a new stadium in town. Then the stadium holds a game with a team from a nearby town, everyone in that town wants to drive to see it, so all the streets in that town are now clogged.
Should the stadium pay the other town money to hold a game because lots of people there drive to watch it? Or should the other town build their roads to meet the demand of their citizens? I think the second option is more reasonable, the problem was that there wasn't enough road for the citizens in general, not that there wasn't enough road for them to watch this particular game.
Envision every driver owning a semi-truck made by the company. It's not a single company blocking the streets; it's the drivers who purchased the company's very large vehicles and who also paid for street access.
It's more of a "false advertising" arrangement than that, though.
If the ISP is selling me 1Gbps access to specific internet content only and I purchased the plan I use based on that knowledge[0] then charging me a toll for things that are outside of that specific internet content would be acceptable. But I didn't. I signed up for service that was limited only by speed, not consumption.
It's a little more evil than that, though. In a consumption/speed model, the two work against each other. I've paid a premium for 1Gps service over 300Mbps which results in me hitting that consumption barrier much faster, so I've paid for the privilege of getting hit with more "tolls". That's a hell of a profitable business model if you can get customers to actually do business with you.
Using a different analogy, the car wash company by ,y father's house offered $3/use car washes or a $30/month plan with hand-towel drying. It's a profitable arrangement for the car wash because most people won't wash their car ten times in a month. My dad, however, owned a company that sold products to automotive manufacturing plants and placed an unusually high value on having a very clean car. He took his car in for a wash every morning and many afternoons. That was the point of buying the $30 plan. For the car wash company to turn around and say "You're using your unlimited service so much that you're negatively affecting access to the car wash for our more profitable customers" would be illegal. That's not to say it doesn't happen all the time in other places, but it also doesn't mean we should allow it.
[0] And it's not OK for that to be buried in paragraph 11 of subparagraph a) in the user agreement as "abuse of network services".
This already happens. The big trucks are UPS and FedEx and Amazon. This companies already pay to maintain their trucks, which includes taxes for road maintenance through gas tax and registration fees.
Those trucks are only on the road because all the people in the houses keep ordering stuff online.
Should FedEx/UPS/Amazon have to pay an extra tax just because they are popular with the people? The homeowners already pay taxes to maintain the roads to their houses. Why should the city get extra from the trucking companies?
A lot of commentors here are defending that netflix shouldn't pay data charges.
I don't know how internet prices works for big players, but from the perspective of a little guy, it seems kinda reasonable that netlflix pay, doesn't it?
I mean, why am I paying for bandwidth on AWS? Shouldn't that be paid for by the downloading party?
If I download something from my neighbors computer over a p2p app, then we both still pay for our internet connection, and depending on the contract, we pay by the byte.
That, though, is exactly how it works. Netflix pays for their traffic and infrastructure, the SK clients pay for access to internet for a given bandwidth and SK pays for the infrastructure and interconnections needed to give their clients what they paid for.
> why am I paying for bandwidth on AWS? Shouldn't that be paid for by the downloading party?
you're paying for it because it costs money to provide bandwidth on both ends, and AWS is just shunting the cost onto you itemized.
Netflix, believe it or not, also charges for the bandwidth it uses, but it hides it unitemized into the subscription fee, and in sufficient quantities that when averaged across all it's subscribers, they come out at least breakeven.
Somewhat unrelated, but your point about costs coming out breakeven when averaged over all users made me realize that this move by SK makes as much sense as Netflix suing an ISP because their customers consume a disproportionately large amount of Netflix' bandwidth. The ISP's role in facilitating the usage of Netflix bandwidth would be the same as Netflix faciliting use of SK's bandwidth in this case.
Because when you signed up for services on AWS, you agreed to pay for egress and ingress bandwidth delivered at very high speeds.
When I signed up for my ISP, I agreed to pay for a rate of speed that was sold to me as "unlimited" with some restrictions, but none that even came close to resembling "don't download video too much".
It also ignores the value Netflix provides. My parents chose to buy faster internet service than they otherwise would have purchased because of Netflix. If Netflix was not in the picture, they would have purchased the cheapest plan available. It's not a fairness failure, it's a marketing failure.
You pay for bandwidth on AWS. In the Netflix scenario, the ISPs sitting between your service and your customers would be demanding that you pay them for using their bandwidth.
The only difference between you and Netflix is that they are AWS with all of the interconnect and overhead costs, while you get to pay a third party to abstract all of that complexity away for what is essentially a management fee.
Netflix DO pay data charges, on hosting their own infrastructure. Probably unthinkable amounts. It certainly isn't up to the hoster to cover an ISPs costs for the people access their service.
SK Broadband's customers should be suing.
The only guilt of Netflix (besides selling down-level show at rip-off price) is what their action helped to reveal SK Broadband can not actually service their customers with agreed bandwidth (and no, its ISP's sole responsibility to pay their uplink for required bandwidth). I'm pretty sure the problem was well known to geek minority, just this streaming spike uncovered the problem to general public.
It's strange that SK Broadband would go to so much trouble rather than be happy for the profit opportunity that Netflix brings since their content is cacheable.
In Nigeria which arguably is more broadband-challenged than Korea, it took me only a few weeks to ship in a Netflix OpenConnect appliance and integrate into the network at a local fiber company where I did some work last year. All we had to do was agree with Netflix that we would keep it powered up and cover the cache-fill bandwidth costs.
Net results: Today, that ISP uses less than 2Gbps of Internet bandwidth for less than 8 hours daily to update their cached content which they deliver to thousands of Netflix users at wire speeds.
Literally you're spending less than 2Gbps to acquire popular content and selling more than 10Gbps to your users during peak periods (if your local infrastructure can support it).
Sk broadband already got paid from their customers. Netflix is merely incurring more costs for them, which is why they are suing in court in the first place!
Sk broadband would prefer that their customers don't have netflix, but only browse text websites!
I worked in pricing and I always framed it as a scale with "Greed" on the left and "Customer Service" on the right and the 2 ideas being in a tug of war.
Greed wants to charge customers infinite money and provide 0 product in exchange for it. Customer Service wants to give away everything for free to as many people as possible.
To be a real, viable business you need to find that sweet spot where neither of the two makes too many of the decisions.
> Netflix's data traffic handled by SK jumped 24 times from May 2018 to 1.2 trillion bits of data processed per second
What an odd way of representing data transfer bandwidth: Was it so hard to just say 1.2 Tbps (or better 150 GB/s)?
Also, maybe an ignorant question from my side: Is that bandwidth such a big deal at nation-scales? It feels like barely 1000 homes maxing out the fiber usage. All submarine cables being deployed today are already in the 100+ Tbps range, aren't they?
Netflix pays for the packets that go in and out of its network. SK’s customers pay for the packets that go in and out of their home networks. So what am I missing? That’s sounds like the way things are supposed to work.
When I read this description of SK’s complaint it appears they are unhappy with the deals they struck with their customers. Nor in other words they are unhappy their customers are actually using the product.
I think it's fairly simple: in the original arrangement where ISPs provided access to content and that content was not video, ISPs were making out like bandits offering unlimited service at ridiculous speeds while not having to upgrade their own connections since customer demand was lower. Suddenly content comes along that increases customer demands on their "unlimited service" and instead of the ISP deciding correct the capacity imbalance (and accept that they "done screwed up"), they cry "unfair" and run to the courts.
Back when peering agreements were in place it made sense for everyone involved; ISPs get great content they don't have to provide and/or pay for, content providers get access to customers. Now, however, ISPs are finding out that these agreements are very costly for them, they can't turn to their customers for more money due to competition and they can't just block Netflix because those same customers would leave for another ISP so they turn to the courts to repair their mistakes. If they win, Netflix gets screwed, the customer gets screwed and the ISP gets rewarded for their mistakes... that sounds like a heck of a public-private partnership. :)
The Internet is literally a series of erbium doped tubes, analogous to highway traffic.
Videos of cats at high enough bandwidth can melt optical interconnects.
For the same reason truckers pay some tax for brutalizing the roads, giant content providers beaming data through these systems must be taxed lest the Internet infrastructure be destroyed and the cost only be burdened by general taxpayers instead of users. (because Netflix would just charge users more, but in the meantime they will fight to have other non-user taxpayers maintain the infrastructure)
No, it isn't. Cars can't spawn clones as they travel down a branching set of roads.
But IP Multicast can do this, which is how IPTV works on the ISP level. And multicast is basically emulated by edge distribution boxes so that the traffic moving into the ISP can be multiplied to multiple customers.
I think in your analogy netflix is more the shop that those truckers are delivering goods to.
So if you want to be more accurate, the proposed payments won't be like a tax on truckers (in this analogy the truckers are the users, or maybe ISPs), rather it'll be a tax on supermarkets and shops that are getting the goods.
I find this thinking just bizarre and displays quite a bit of hubris from the ISPs and/or S. Korean courts.
What's happening here is simple: Netflix is so popular that an ISP dare not block their content. Really, put simply: Netflix costs money to deliver, yes, but it is an expectation of ISP customers that they have access to Netflix. If an ISP were to block/charge more for access to Netflix (assuming that's even legal, but let's just say it is), customers would move to another ISP. Outside of cases where moving to another would be impossible[0], that should be the end of it.
It's interesting to me that the article focuses entirely on the courts getting involved. What's happened is that a business made assumptions about cost (that "unlimited" would mostly be users surfing lower-bandwidth web sites, not delivering video) and the market (all of their competitors likely land somewhere near eachothers' pricing) and discovered that those assumptions were wrong. Instead of either accepting that they are not a commodity that will always chase the lowest price and operating accordingly, they've gone to the courts to have the government require Netflix to reward them for their mistakes.
This idea that as an ISP customer, you "paid for unlimited service at a certain rate" but "don't you dare actually use that unlimited service" is absurd. Watching Netflix, even the vast majority of the day/night, isn't a matter of abuse -- akin to bringing a sleeping bag and living at a 24-hour gym that you have a membership to -- it is what most ISPs are advertising as a reason to subscribe to them. They know they can't turn around and tell their customers (that have choices in service providers) that they're going to charge them for access to something that is central to that service, so they go after Netflix. What if Netflix refuses to pay and just leaves the S. Korean market or decides to say "screw it" and shut the business down (ala Atlas Shrugged)? Would that make S. Korean ISPs service more valuable?
[0] In these cases, they're a monopoly to those customers and should be regulated accordingly; preferably in a manner that brings in an ISP that can handle Netflix traffic without the charges.
242 comments
[ 2.8 ms ] story [ 226 ms ] threadComcast is at $18b in operating profit, with roughly 18% operating income margins. Charter Communications is at $9.6b in operating income, with 19% operating income margins. Those are the two biggest cable companies in the US.
In the early decades of cable, profitability was notoriously flaky for the industry. Enormous amounts of required capital investment, and it took a long time to saturate - scale into - the US market. Now the market is settled and they're printing consistently giant profits.
Interestingly Charter has become a giant, worth $133 billion, and Paul Allen used to own a controlling interest in the early incarnation of it. It was part of his so called wired world collection of investments.
https://www.seattletimes.com/business/economy/paul-allen-cha...
In 2020, KT: 1.18 trillion KRW (1.00bn USD), SKT: 1.34 trillion KRW (1.13bn USD), LG U+: 886 billion KRW (749m USD). Their revenues range from 10 to 25 trillion KRW for the reference.
If my ISP is losing money then I guess they better raise their prices, no?
However, that would leave free content inaccessible, and probably create another wealth divide.
That's leads to a pretty simple conclusion, because Netflix doesn't push any data. I pull data from Netflix. I have already paid my ISP to pull data from anywhere I ask. If my ISP sits down at a negotiating table with Netflix and threatens to disrupt that, they should be laughed out of the room. "Pay me, or else I'm going to break my obligations to an unrelated third party." isn't a move that should have any respect.
But the problem is the people that could get affected by their actions until they stop.
Regardless, Netflix is not pushing content. The consumer is pulling it.
As a one-off, issues like the Squid-Game aren't so bad, but if it's a chronic problem for the ISP then it's because their model of consumer consumption was wrong.
Imagine if the same principles were used in other areas. You iPhone suddenly locking up after using Facebook to much because Facebook wasn’t paying Apple for access to its users.
Pushing implies that data is sent by the server without clients explicitly requesting it.
Pulling is when the client requests data explicitly.
I get notifications pushed to my phone by Apple.
I pull YouTube videos.
YouTube doesn't suddenly decide to "push" random videos to my phone in the middle of the night.
Similarly, Netflix doesn't "push" videos to their customers.
In other words, they will ship some Open Connect Appliance boxes to the ISP.
https://openconnect.netflix.com/en/
And that’s nice for everyone.
Some think it’s unfair that big players like Netflix can do so. But I think them doing that is far far better than the idea of ISPs charging content providers for the bandwidth.
Like other said, the customers of the ISPs are already paying for the bandwidth. Having the ISPs charging Netflix for the bandwidth too, that’s no bueno.
But it's your customers (the end users) generating traffic by watching content on my network (Netflix shows). If your customers can generate more traffic than you can handle, regardless of its a single entity or multiple entities, then that's a contract issue between you and your customers.
So the philosophical question really becomes, what exactly is the end user buying from their ISP?
Wouldn't that also include the bandwidth? Meaning, if the customer is already paying for the bandwidth they will consume, why wouldn't that cover bandwidth usage of Netflix content?
Besides, physical access alone is useless. Unless you're claiming that ISP's have been selling physical access but giving away usage for free? I don't know if that's the intent of your comment, but again it isn't how my ISP words the contract.
I don't think zakki is too far off from the truth. Looking at Spectrum's Residential Internet Services Agreement, physical access might be what it boils down to. It's a little shocking they can have terms that make no guarantee to any amount of bandwidth at all, even on "their network". I suspect other large ISPs in the US use the same terminology.
>Subscriber understands and agrees that Spectrum does not guarantee that any particular amount of bandwidth on the Spectrum network or that any speed or throughput of Subscriber's connection to the Spectrum network will be available to Subscriber.
But I think you're correct. Having a physical connection does you no good if it isn't passing packets.
https://www.spectrum.com/policies/residential-internet-servi...
I bet that wouldn't go over nearly as well as this.
SMS spam, though, had a very clear injury when you were charged for the privilege of receiving a message.
[1]https://support.xbox.com/en-US/help/hardware-network/console...
https://www.fastcompany.com/90580656/comcast-data-cap-remote...
The way I look at it is that my plan is $20 then my data is $1 / 100 MB.
To get significant damage as a single individual, take someone on a traditional satellite connection who can't get some important work done on time because ads put him above his data cap.
Advertisers give money to users give money to content provider
Where money given to the users == money given to the content provider
and the users pay for the service the advertisers are consuming.
Consider a user with a data cap of 1000GB, is metered as consuming say 1200GB of data, of which 200GB is Ads. The user is forced to pay extra for the 200GB to the ISP which the user should not have had to pay if there were no Ads.
Ignoring that ISPs should not meter the service, the above just doesn't seem right. The individual Internet services may be subsidized by Ads, but the user is paying to the ISP. So, in effect a user is paying, just to a different party.
Some lawyer should definitely look into it and see if there is potential for a class-action against Ad vendors.
Edit: I'm sure the flip argument would be that the user requested the Internet services that came with the Ad, so in effect the user is responsible for consuming the data. But that is where I think it becomes interesting and given that IMNAL I don't know what the legal minds think.
I mean there are only a few ISP in every country. SK Broadband isn't small either it is something like 25% of market share. If I remember correctly Korean Telecom was 50% and LG was something like 15%. SK Telecom is also the largest MNO is South Korea. And if you are into hardware you may have at least heard of SK Hynix, world's third largest NAND manufacture.
Normally being a Korean conglomerate or Chaebol I would have thought SK must have some subsidiary competing with Netflix. But this doesn't seems to be the case here.
Netflix made a few big splash with content made specifically for South Korean. Or K-Drama. But I dont have any data to judge whether they are super popular or not.
Yes, from the article:
> The popularity of the hit series "Squid Game" and other offerings have underscored Netflix's status as the country's second-largest data traffic generator after Google's YouTube
By having Open Connect Appliance network gives up ability to traffic engineer Netflix, which may decrease the leverage the network has with its peers or its transit providers.
Fundamentally, unless someone must peer with you (settlement or settlement free) and must peer with you on your terms ( locations, types of connections, etc ) you always want traffic that you and only you can control.
Your guess is indeed correct! Two out of three major ISPs in South Korea operate their own OTT services (Wavve and Seezn). LG U+ is an exception and also the first major ISP to join the Open Connect Appliance program for the obvious reason.
It's impressive how this framing has gotten so much acceptance. The content providers are not generating the traffic, the users are, and they are paying their ISPs to be able to do so.
If I could make one simple change to the home broadband market, it would be insisting on a standardized big number consumer level SLA. I certainly don't expect residential rates to cover the same 99.97% or worst case minimum that business class does, that's a major part of what higher end connections are paying for. But whatever they do offer should be transparent, fixed for contract period and measurable. ISPs should by law be required to spell out not just the top line bandwidth level, but also 75%/95% guarantees over a standard time (day/month maybe) and what the worst case guaranteed minimum is. This wouldn't be some big centralized bureaucratic decision about implementations or what numbers have to be hit, but rather merely make sure the market has clear information so customers can compare and ISPs could be held to their promises. Energy Star consumer information might be some guide to layout, but I think something like that would be low hanging fruit.
Also to be clear, a distinction can and should be made for intra-network vs inter-network, and both should be accounted for. Obviously an ISP has the most control and is the most responsible for how traffic traverses their own network to their edges, but ISPs do have some responsibility (affected by size/location) for their peering agreements too. If they're purposefully choking that it's on them, though small/remote ISPs may be at a disadvantage in negotiations there which should ideally have some eyes on it.
I have a gigabit connection and the contract specifically states that they guarantee a maximum throughput of 400Mbit at peak times.
It says nothing, there are two weasel word exceptions in there.
So it could save you nothing, or it could save you 15%, or even more. Basically we have no idea what it'll save you, if anything.
Otherwise 0Mbit is consistent with the terms of the contract, which is probably exactly what the ISP was going for.
It does highlight a philosophical question I raised in another comment. What exactly are consumers purchasing from last mile ISPs? Is it 1Gb to anywhere on the providers network? 1Gb to the nearest CDN/Internet exchange?
I don't think there is a good answer for this. But it's important to think about when we talk about ISPs reaching their hand towards large content providers.
this is beyond flawed, unless you consider all possible bandwidth, not just Netflix being slow. The latter can be entirely an issue on Netflix side.
I guess that's the thing with Netflix though, it's torrent-like traffic, but legal, so there's no excuse to stop it and they have to come up with something. "oh they're generating explosive traffic that's not good"
First, they aren't being paid for unlimited. Do you think that $39.99 or $199.99 per month can cover the cost of transferring unlimited amount of data to another continent? Of course, it doesn't. Somebody pays $39.99 to download 10 emails, another customer runs BitTorrent 24 by 7 consuming TB's of bandwidth for the same price.
Second, they don't want to sell "unlimited" - they could make more money by selling differentiated services.
I'm not sure if the lawmakers are ignorant, bought and paid for, or a little of both but it would be refreshing if one of them took 30 seconds and applied a little critical thinking to the situation.
Wouldn't that require them to MITM the Netflix traffic, which I would hope is under TLS?
I believe YouTube has the same type of program.
Netflix calls it their Open Connect Appliance: https://openconnect.netflix.com/en/appliances/
That server would, of course, have a connection to the Internet, they push logs to AWS and get updates from Netflix's authoritative servers in Netflix's other datacenters, but they also serve HTTPS connections to thousands of ISP users; the upstream traffic would be minimal.
https://openconnect.netflix.com/en/
This problem might also be an interesting use case for Signed HTTP Exchanges. Sign the video files, then push them to the caching server. The caching server then never possesses the private keys for the connection and also cannot modify the content.
What are you going to achieve by stealing the certificate for edge0.sktelecom.geo.netflixcdn.com from the box you already have access to?
There’s no reason to put *.netflix.com certs on the edge caches.
Versus, what's the attack? The ISP with physical access to the box can pull can screw with the cached video data somehow? Or MitM and know what Netflix videos their users are watching?
Also, the less said about SK's neighbors (Japan, for instance) the better.
"But SK Broadband demanded that Netflix pay network usage fees on top of installing those caches. The streaming service sued the telco, and that case is pending — Netflix told us that it would continue negotiations even as it made its case in court."
Alternatively, they would peer with SK Telecom but "SK was demanding large content providers money for peering with them."
https://www.medianama.com/2020/08/223-net-neutrality-south-k...
There is nothing illegal about torrent traffic any more than any other form of traffic. "It's streaming-like, but legal." "It's VPN-like, but legal." "It's cyptolocker-like, but legal." "It's SSL-like, but legal." "It's Tor-like, but legal."
I am surprised that ISPs haven’t started cracking down on WfH, and start saying it violates their TOC because this is a business use. Public uproar would be immense.
You listed the reason why yourself, you shouldn't be surprised. Maybe -after- the pandemic, when WfH people can be painted as "tech elite" or something, but not while everyone wants to WfH and it's a mixed bag who can.
Actual, I might be wrong here with how contact law is handled ... anyone?
A related consideration: when we moved into our terraced house the land was under leasehold and the lease specified 'no business use'. We bought the freehold some time back, but I've often wondered what my position might otherwise have been. Presumably breech of the lease conditions would be a tort against the landlord and so I'd have been fine as long as they didn't sue.
No, it's not unreasonable. They sell consumers unlimited high bandwidth connections and at no point are they told they can't actually use the service they're paying for. What's unreasonable is cutting off paying consumers and suing streaming services when the company's dumb business model starts failing because of their own invalid assumptions. If oversubscribing is such a problem, they should stop doing it instead of blaming consumers for overusing their low capacity network.
Lashing out at Netflix and bringing them to court for their own failings is such an absurd course of action. They should try upgrading their networks instead so that their service stops sucking.
Since when is ANY traffic illegal on the Net? Is the idea of an open (for everyone) data "highway" already death in our heads?
It's even more impressive how the average HN reader can't even comprehend this comment (or can, but somehow manages to insist on their wrong, socialist views) and continues to argue in favor of "network neutrality", even though no two users or service providers are the same.
If Netflix should pay $10 mil a month to Korea's Internet providers, why not HN? If HN should pay less than Netflix, why should the user who only visits HN pay the same Internet access fee as the one who constantly watches Netflix?
https://bgr.com/general/netflix-vs-verizon-network/
Now Netflix blocks Hurricane Electric endpoints and FiOS still doesn't support IPv6. We're nearing the 10 year anniversary of that "coming soon" FAQ entry on their webpage.
Online ads and data vacuuming is often presented in a similar fashion. Your browser goes to the advertiser and requests the ad. The advertiser asks for your data. Your browser sends the data to the advertiser. And the way we frame this is that advertisers are stealing our data.
Once a system is opaque and automated enough it doesn't seem to matter how it really works. Netflix is a big entity, therefore we blame Netflix. If Netflix didn't exist then the traffic surge wouldn't exist, therefore Netflix should pay.
Netflix/youtube cases it's the content providers that server the traffic, so most of it is generated by them. The client requests are tiny in comparison.
/You pay datacenter for bandwith etc.
Person pulls your website
/Person pays cablecompany Z for bandwith
Now you have to pay Cablecompany Z for the data.
Admittedly the article is light on the details, yet I'd presume the line has been created exclusively for Netflix, there is no info if there was an agreement between the parties. There is no reference to the Seoul District Court decision, either.
"Internet"/data between major tube providers (ISP) is very different than paying cable company X for end user access. Some just have agreements to carry each other traffics and some pay to carry their traffic[0]:
[0]: https://en.wikipedia.org/wiki/Tier_1_network
When there are alternatives, ISPs would bend over backwards for what customers want.
That are not colluding like a cartel, of course.
Huh?? I am not serving terabytes of movies from my home connection. Netflix is.
Is the ISP mad that more users are using more of their promised service allowance? Too bad, maybe they shouldn't oversell their networks.
https://www.streamingmediablog.com/2014/02/heres-comcast-net...
South Korea is well known for having fast internet, but the internet is only fast within the small country. International connectivity has always been crappy, and getting crappier every year as telcos focus on legal avenues instead of actually laying enough undersea cables to meet the demand. In the past, most Koreans only visited Korean sites because of the language barrier; that's no longer true. But it's going to take a lot of time and money to remedy years of neglect in international connectivity. Especially SK Broadband, who IMO has been the most complacent in this department among the Big 3 telcos, precisely because of their huge domestic market share. Where is the money going to come from? Obviously not by cutting executive bonuses. SK Broadband seems to think that they can extort it out of Google and Netflix who, despite their global prominence, are relatively defenseless in Korean courts.
IIRC there were some companies that got a slap on the wrist (it was something like Uber or a food delivery company) for charging both consumers and producers a commission, which is illegal (I don't remember the legalese term).
The rest of the increase would go to offsetting the additional burden on Netflix that this will cause by forcing them to treat S.Korea differently than everyone else.
I'm honestly surprised that any corporations are paying that there, but if the government is going to requirement, there's not a lot you can do about it. It's a very bad precedent and I don't see it actually going well for S. Korea. Putting up extra barriers is a way to keep companies out. We usually call it a "tariff", but that's generally for physical goods. And the goal is often to prevent as many of those goods from being imported.
What would S.Korea's internet look like if even 1 of the major players decided to just block the country instead of paying the fees?
It benefits Netflix in the long term as it raises barriers to entry for disruptive competitors.
If, in order to reach an ISP's customers, you have to pay the ISP for the privilege, then creating new startups is harder and the incumbents can just price it in.
I wouldn't be at all surprised if this became more common elsewhere, as everyone involved benefits. Except us, the customers, of course, but we don't get a seat at this table and the people who should be fighting for us are taking bribes (sorry "campaign donations").
Content is king and competitors are already priced out long before minor squabbles over bandwidth.
It's like "free guy" (spoiler boring) tries to be a critique of video Game violence by positing the existence of another gaming reality where people are nice to each other. But this other reality is just a Macguffin. We come to watch the violence and mayhem and we come to watch squid game for the same reason.
It is better to own the fact that watching torture porn gives you the jollies rather than posturing that it gives you membership of some anti capalist intelligensia.
It's just Romans and Christians and Lions. The crowd loves it because people love blood.
The comment you’re replying to says the show is harrowing and a critique. There’s no “because” or apologia.
Considering your comment seems to be targeted at pseudo-intellectualism the irony here is simply off the charts.
> capitalism
Too bad they don't want to watch award winning Kim and Beautiful in sustainable 320×200 format from the North :)
I just saw the final episode last night, and I don't remember seeing many flashbacks other than at the start of E01.
I admit it is slower moving than I'd like, and some of the episodes had way to much filler and not enough plot arcs, but I seriously do not remember seeing flashbacks.
Do you have some exclusive access to a version of Squid Game that we all don't?
If I were the judge, I would ask Netflix to stream low bandwidth version of the content so it won't impact SK Broadband.
AND, tell BOTH parties that they must put announcement in their respective website, "SK Broadband provides you internet at a cost (as per their claim). To enjoy Full HD of this content, pay extra to SK Broadband".
See which party the customers will be angry to.
I wonder if we'll ever see a predictive home-caching system; if I watch the first three episodes of squid game in one night, it might be sensible to have my TV download the rest of the series overnight.
Short version: our Netflix is just fine now. There were just a couple rocky weeks/months at the start. :)
Instead of using the internet its also feasible to use the DVB network like this. they could even broadcast a copy long before release so its guaranteed to be ready for offline watching when released. i am almost certain Sky Germany does something similar...
The public street network is in large parts paid for by taxes, so for everyone access to the network is already payed for.
If I now envision a single company blocking all streets with its giant truck fleet, I could completely understand a motion to introduce a new form of toll to be payed by such companies to avoid having to rise taxes.
That toll would most likely just increase the price of whatever said company is selling.
In the road example, it's more like if every person paid for the ability to put one truck on the road and everyone delegated their truck space to a shipping company who now has millions of trucks hogging the roads. Yes, they're using tons of infrastructure but their usage has been paid for already.
Should the stadium pay the other town money to hold a game because lots of people there drive to watch it? Or should the other town build their roads to meet the demand of their citizens? I think the second option is more reasonable, the problem was that there wasn't enough road for the citizens in general, not that there wasn't enough road for them to watch this particular game.
If the ISP is selling me 1Gbps access to specific internet content only and I purchased the plan I use based on that knowledge[0] then charging me a toll for things that are outside of that specific internet content would be acceptable. But I didn't. I signed up for service that was limited only by speed, not consumption.
It's a little more evil than that, though. In a consumption/speed model, the two work against each other. I've paid a premium for 1Gps service over 300Mbps which results in me hitting that consumption barrier much faster, so I've paid for the privilege of getting hit with more "tolls". That's a hell of a profitable business model if you can get customers to actually do business with you.
Using a different analogy, the car wash company by ,y father's house offered $3/use car washes or a $30/month plan with hand-towel drying. It's a profitable arrangement for the car wash because most people won't wash their car ten times in a month. My dad, however, owned a company that sold products to automotive manufacturing plants and placed an unusually high value on having a very clean car. He took his car in for a wash every morning and many afternoons. That was the point of buying the $30 plan. For the car wash company to turn around and say "You're using your unlimited service so much that you're negatively affecting access to the car wash for our more profitable customers" would be illegal. That's not to say it doesn't happen all the time in other places, but it also doesn't mean we should allow it.
[0] And it's not OK for that to be buried in paragraph 11 of subparagraph a) in the user agreement as "abuse of network services".
Those trucks are only on the road because all the people in the houses keep ordering stuff online.
Should FedEx/UPS/Amazon have to pay an extra tax just because they are popular with the people? The homeowners already pay taxes to maintain the roads to their houses. Why should the city get extra from the trucking companies?
I don't know how internet prices works for big players, but from the perspective of a little guy, it seems kinda reasonable that netlflix pay, doesn't it?
I mean, why am I paying for bandwidth on AWS? Shouldn't that be paid for by the downloading party?
If I download something from my neighbors computer over a p2p app, then we both still pay for our internet connection, and depending on the contract, we pay by the byte.
That’s a great question; it’s because cloud providers have conditioned you to believe their rent-seeking is necessary.
See Cloudflare R2 vs AWS S3: https://news.ycombinator.com/item?id=28702997
Netflix doesn't get free interconnections.
you're paying for it because it costs money to provide bandwidth on both ends, and AWS is just shunting the cost onto you itemized.
Netflix, believe it or not, also charges for the bandwidth it uses, but it hides it unitemized into the subscription fee, and in sufficient quantities that when averaged across all it's subscribers, they come out at least breakeven.
When I signed up for my ISP, I agreed to pay for a rate of speed that was sold to me as "unlimited" with some restrictions, but none that even came close to resembling "don't download video too much".
It also ignores the value Netflix provides. My parents chose to buy faster internet service than they otherwise would have purchased because of Netflix. If Netflix was not in the picture, they would have purchased the cheapest plan available. It's not a fairness failure, it's a marketing failure.
The only difference between you and Netflix is that they are AWS with all of the interconnect and overhead costs, while you get to pay a third party to abstract all of that complexity away for what is essentially a management fee.
In Nigeria which arguably is more broadband-challenged than Korea, it took me only a few weeks to ship in a Netflix OpenConnect appliance and integrate into the network at a local fiber company where I did some work last year. All we had to do was agree with Netflix that we would keep it powered up and cover the cache-fill bandwidth costs.
Net results: Today, that ISP uses less than 2Gbps of Internet bandwidth for less than 8 hours daily to update their cached content which they deliver to thousands of Netflix users at wire speeds. Literally you're spending less than 2Gbps to acquire popular content and selling more than 10Gbps to your users during peak periods (if your local infrastructure can support it).
https://openconnect.netflix.com/en/
Sk broadband already got paid from their customers. Netflix is merely incurring more costs for them, which is why they are suing in court in the first place!
Sk broadband would prefer that their customers don't have netflix, but only browse text websites!
Greed wants to charge customers infinite money and provide 0 product in exchange for it. Customer Service wants to give away everything for free to as many people as possible.
To be a real, viable business you need to find that sweet spot where neither of the two makes too many of the decisions.
What an odd way of representing data transfer bandwidth: Was it so hard to just say 1.2 Tbps (or better 150 GB/s)?
Also, maybe an ignorant question from my side: Is that bandwidth such a big deal at nation-scales? It feels like barely 1000 homes maxing out the fiber usage. All submarine cables being deployed today are already in the 100+ Tbps range, aren't they?
When I read this description of SK’s complaint it appears they are unhappy with the deals they struck with their customers. Nor in other words they are unhappy their customers are actually using the product.
Back when peering agreements were in place it made sense for everyone involved; ISPs get great content they don't have to provide and/or pay for, content providers get access to customers. Now, however, ISPs are finding out that these agreements are very costly for them, they can't turn to their customers for more money due to competition and they can't just block Netflix because those same customers would leave for another ISP so they turn to the courts to repair their mistakes. If they win, Netflix gets screwed, the customer gets screwed and the ISP gets rewarded for their mistakes... that sounds like a heck of a public-private partnership. :)
Videos of cats at high enough bandwidth can melt optical interconnects.
For the same reason truckers pay some tax for brutalizing the roads, giant content providers beaming data through these systems must be taxed lest the Internet infrastructure be destroyed and the cost only be burdened by general taxpayers instead of users. (because Netflix would just charge users more, but in the meantime they will fight to have other non-user taxpayers maintain the infrastructure)
No, it isn't. Cars can't spawn clones as they travel down a branching set of roads.
But IP Multicast can do this, which is how IPTV works on the ISP level. And multicast is basically emulated by edge distribution boxes so that the traffic moving into the ISP can be multiplied to multiple customers.
[0] https://en.wikipedia.org/wiki/Series_of_tubes
So if you want to be more accurate, the proposed payments won't be like a tax on truckers (in this analogy the truckers are the users, or maybe ISPs), rather it'll be a tax on supermarkets and shops that are getting the goods.
What's happening here is simple: Netflix is so popular that an ISP dare not block their content. Really, put simply: Netflix costs money to deliver, yes, but it is an expectation of ISP customers that they have access to Netflix. If an ISP were to block/charge more for access to Netflix (assuming that's even legal, but let's just say it is), customers would move to another ISP. Outside of cases where moving to another would be impossible[0], that should be the end of it.
It's interesting to me that the article focuses entirely on the courts getting involved. What's happened is that a business made assumptions about cost (that "unlimited" would mostly be users surfing lower-bandwidth web sites, not delivering video) and the market (all of their competitors likely land somewhere near eachothers' pricing) and discovered that those assumptions were wrong. Instead of either accepting that they are not a commodity that will always chase the lowest price and operating accordingly, they've gone to the courts to have the government require Netflix to reward them for their mistakes.
This idea that as an ISP customer, you "paid for unlimited service at a certain rate" but "don't you dare actually use that unlimited service" is absurd. Watching Netflix, even the vast majority of the day/night, isn't a matter of abuse -- akin to bringing a sleeping bag and living at a 24-hour gym that you have a membership to -- it is what most ISPs are advertising as a reason to subscribe to them. They know they can't turn around and tell their customers (that have choices in service providers) that they're going to charge them for access to something that is central to that service, so they go after Netflix. What if Netflix refuses to pay and just leaves the S. Korean market or decides to say "screw it" and shut the business down (ala Atlas Shrugged)? Would that make S. Korean ISPs service more valuable?
[0] In these cases, they're a monopoly to those customers and should be regulated accordingly; preferably in a manner that brings in an ISP that can handle Netflix traffic without the charges.