Netflix are right, SK Broadband want to be paid twice for the bandwidth their customers use, once from their customer and again from Netflix.

SK want to charge both ends of the data connection their customers and who they transfer data from/to even though the other end also has to pay their own carrier.

So, I assume that Netflix’s carrier can charge SK for all the bandwidth SK’s customers use downloading from Netflix?

Trying to see it from the other side, perhaps what SK broadband are saying is that Netflix should pay for their bandwidth so SK Broadband's customers don't have to, especially those who don't use Netflix.

If this were the UK I'd have my doubts that the savings would be passed to the customers but perhaps the South Korean market works differently.

SK are saying that they have sold their customers download capacity that their network can't support when a lot of their customers decide to use that download capacity.

It is irrelevant where the packets are coming from, SK have sold their customers a service that they can't provide when a large number of customers use that service.

That's certainly one way to look at it but what I was trying to suggest is that the South Korean market sees things differently. I don't know, the nearest thing which comes to mind is the banking system in the US. I heard from colleagues who moved there that dealing with banks from another state is painful, which seems odd when compared to Europe where I can transfer money to different countries with ease.

In practical terms sending money between banks across regional or country boundaries is technically simple as proven here but for cultural and historical reasons I hear it's difficult in the US. In practical terms 1 packet is the same as another but perhaps in South Korea there is some cultural/historical reason why it doesn't work like that?

While I've visited South Korea I've never used the Internet there and my Korean is only really good enough to ask basic directions and on a good day order some food :-) So I don't really know how things stand over there with regards to ISPs.

>It is irrelevant where the packets are coming from, SK have sold their customers a service that they can't provide when a large number of customers use that service.

It looks like you are coming from the fabulous BT world where one is guaranteed 67Mbps on average or get your £20 back. They never failed to deliver, didn't they?

SK Broadband sell plans like "up to 10Gbps for £32/mo". Unlike BT who absolutely always respect the minimum bandwidth guarantee, SK Broadband do not promise their customers that they always get 10Gbps or whatever (1Gbps seems to be the minimum they offer) so your point suggesting the breach of contract from the ISP does not seem to be valid.

When SK Broadband's 8+ mln subscribed households start watching Netflix in 4K while playing games and video calling friends, the Netflix traffic might look impressive enough and growing to start asking if anything needs to be done about it. The fact that Netflix is SK Broadband's direct competitor for IPTV does not help either (to date, Netflix user base is South Korea seems to be only half of SK Broadband's).

What we seem to witness here is a large South Korean company trying to stop a large US company from taking hold of their South Korean user base by suing them in a South Korean court of law—or at least charging the hell out of them for doing so. Take it easy, no customers were harmed in the making of this drama.

They must be massively oversubscribed if they consider 1 Gbps to a "minimum" yet Netflix which is maybe 20 Mbps per stream if everyone is watching 4K is causing them problems. Are their customers streaming on 50 TVs at once in every household?

I agree and we keep getting told - by Openreach Bashers - that you can get ubiquitous full fibre broadband to the most rural Kimchi farm is the deepest South Korean sticks. Bandwidth there should not be an issue.

… unless this is just Openreach bashing fakery… as Akami global broadband metrics keep showing.

Notes for Netflux:

Cue Netflix - a 5 second or less advertorial before every show starts telling them that they should migrate away from SKT?

Email all SK customers who are with SKT tell them of a special offer for another SK ISP (who has openconnect) with an introductory deal and say as of the 1st of Jan all Netflix customers who are still with SKT will pay an extra 25% to pay SKT for bandwidth charges (if they lose the court case) and give them the phone number for SKT if they do believe that it is fair and do not wish this to happen?

Failing that pay me for the idea of a comedy series giving only plausible deniability that it's about SKT making them out to be grasping greedy monsters. You could do all sorts of office meeting situations making them look very evil and involved in anything that is frowned on in SK. Of course the show would be prefaced with a statement like "This is a work of fiction, any resemblance to any company or any person dead or alive is purely coincidental.". Fictional company motto - You will pay for that!

That's a very good idea if the goal is to have Netflix investigated for anti-competitive practices. South Korea made Apple and Google comply with the local antitrust law, Netflix should be easier.

is inversely proportional to the quality of its Content.

Meh. The writing is on the wall.

The entire world has HAD it with American megacorps reaping profits from their nations, paying little to nothing in taxes and infrastructure/service costs, and leave the local infrastructure providers with the task of delivering sufficient bandwidth for their poorly written network congestion producing products.

That 15% minimum tax around the world? That's just the first shot in a salvo of a war against the megacorps...

The network is just a pipe.

SK Telecom seems to have forgotten that simple fact.

People only pay for the pipe because of the content they want to consume - not because the pipe is so wonderful.

Pipes can be substituted (fairly) easily and SK Telecom has about 15% market share in the fixed line broadband market in SK (21.5 million households, suggesting about 3m connections).

Netflix have about 3.8m subscribers in SK altogether.

The network provider will 'lose' this battle - they will either back down or lose end customers.

So, this Netflix Open Connect content delivery that saves ISPs so much bandwidth. How does it work? Why would Netflix offer to supply and install it for "free"? What do they get out of it? Is it purely to off the Netflix customers a better service to aid in selling more subs? Or is it a closed "black box" over which the ISP has no control while ot slurps up data the ISP may not want to give away?

Is there a catch? If not, is the potential for a catch there once the ISP is grabbed by the balls?

Cynical? Moi? Naaaaaa....yeah.

I would guess for the following reasons (no inside knowledge here)

This will most likely be a caching server, I.e. first time someone watches a program it will take a copy then the next person who watches the same will be fed off the local server.

The free option is easy by only sending the program once to the caching server netflix save on their own transit costs. And it means that for popular programs there is less network congestion.

And finally if this box was a bad thing why are there over a thousand ISPs who have installed it?

If the ISP host the caching server, then doesn't that mean they still have to deliver all the content, unicast, to consumers?

It might save 95% of traffic at the Netflix end, but probably not at the (downstream) ISP side.

Maybe they can install a caching server in each neighbourhood though, at great expense, with the fees paid to Netflix.

I don't think this is a clear-cut case, I can see arguments on both sides.

"If the ISP host the caching server, then doesn't that mean they still have to deliver all the content, unicast, to consumers?"

That's literally what an ISP is paid to do.

If the ISP host the caching server, then doesn't that mean they still have to deliver all the content, unicast, to consumers?

Yep, and the access portion of a network is the expensive bit, ie installing capacity to deliver a couple of million decent broadband connections.

It might save 95% of traffic at the Netflix end, but probably not at the (downstream) ISP side.

Yep. It's where Netflix is being rather disingeneous. So..

Netflix builds out it's own backbone to popular exchange points, and offers peering there. So cost to Netflix is renting wavelengths or fibre to those points, which on a $$ per Gbps is less than building say a 1Gbps network to a couple of million subscribers.

Where Netflix isn't present, it buys transit to reach it's customers. Transit ISP charges Netflix. As Netflix isn't present in SK, SK ISPs either have to pay for capacity to an exchange point where they are. Alternatively, they'd have to pay a transit ISP.. Usually the one that's carrying Netflix. So that transit ISP already 'double dips' and charges both ends of the connection. If traffic levels increase to Netflix, the SK ISP's costs increase to deliver Netflix's traffic.

If the SK ISP installs a Netflix cache, they can avoid costs of carrying Netflix's traffic either via their peering connections, or their transit provider. But Netflix also saves money, ie it carries less traffic on it's own network, or via it's transit provider. The SK ISP still has to pay for capacity to deliver Netflix's content across it's access network to Netflix (and the ISP) suscribers.

So the '95% cost saving' is probably skewed towards Netflix's own delivery cost, not the ISP.

Maybe they can install a caching server in each neighbourhood though, at great expense, with the fees paid to Netflix.

Sure, but there's a cost in that and also potential operating headaches. One issue is that AFAIK, the caches don't really allow the ISP much control over the traffic. So could be fun figuring out the routing, especially as a lot of access networks are pretty much Ethernet because the cost of delivering LAN services via tech like GPON is a lot less than installing and managing a lot of large (and expensive) routers.

But Netflix's position is pretty clear. It doesn't want to pay for delivering it's customer's traffic, so ISP's only option is to increase costs to their subscribers to pay for the impact of Netflix (and other OTT content providers) traffic.

Yes, the ISP has to pay for the bandwidth to their users. However, that's the inexpensive bit (they're moving data across the wires they already built and paid for). The expensive bit for them on an ongoing basis is peering with another network, so if they cache the data in their network, they don't have to pay so much in peering. That's how the ISP can benefit from having Netflix's box on their network, although if they don't think it will help, they're free to not install it and move traffic like they do with everyone else.

"The free option is easy by only sending the program once to the caching server netflix save on their own transit costs. And it means that for popular programs there is less network congestion."

Yeah that's always been my understanding. Plus, if the content is being served direct from your ISP you're likely to see much faster and better quality service as an end user, which looks good for the end user. So both Netflix and the ISP get lower transit costs and the end user gets a better experience. A win, win, win for everyone.

"This will most likely be a caching server, I.e. first time someone watches a program it will take a copy then the next person who watches the same will be fed off the local server."

From looking into it, it seems that Netflix populate the caches off-peak and part of the requirement is a minimum speed rating for the cache server to talk to Netflix to guarantee Netflix data can get through. The type and size of caching server will depend on the ISP and their Netflix traffic levels. So I'd guess that Netflix must, at the very least, be tracking what is going through each ISPs caches and keeping them "topped up" with likely shows based on viewing habits. Of course, Netflix have precise details of what each customer watches and when and almost certainly know which ISP their customer is using so that may be moot anyway.

Why would Netflix offer to supply and install it for "free"?

Presumably to mitigate the sort of thing which is happening in SK, where an ISP asserts the content provider and their subscribers are imposing undue costs upon them.

It could even be taken as an indication that Netflix are aware of the adverse consequences of what they do.

But it's not as simple as one side is to blame the other being blameless.

Take an analogy of a government deciding to give everyone who commutes to work free cash, where the infrastructure for commuting isn't suitable for what is being encouraged. The infrastructure guys will take a beating, be blamed for not being adequate, but they did not create the situation they find themselves in.

Or, perhaps, people being forced into having electric vehicles without enough capacity to charge them.

Yes the problem is that the traffic is clogging up the interconnects to the rest of the internet, and a local server hosted in SK's network would solve that.

The economics behind Content delivery networks (CDNs) are quite simple:

The content provider pays for peering bandwidth when the destinations (Netflix subscribers in this instance) 'sit' on another ISP's network. That costs money, and as someone else said earlier, as on demand video is streamed (unicast) rather than broadcast or multicast, every active video stream consumes bandwidth - and in this case peering bandwidth.

At some point (scale) it becomes cheaper to invest in infrastructure which hosts content closer to the subscribers, thus minimising the ever growing peering bandwidth costs. Hence the rise of content delivery networks. Updates are pushed from a core to the edge nodes once, and the edge nodes then serve the streams in their locality.

It looks like SK telecom is actually concerned about the economics of their own backhaul networks and potentially even local access (something that CDNs don't address) - and in that case Netflix, YouTube, Prime and other (video) services being consumed more and more may require investment in greater bandwidth and less contention than they have previously got away with. Netflix, in that instance, may just be the biggest target for SK Telecom to address their issue.

SK Telecom seem to be caught between a rock and a hard place - price competition for broadband providers on the one hand and rising network costs and required investment on the other. Kicking Netflix off may fix the network capacity issues they seem to experience, but may lead to a customer exodus at the same time.

>So, this Netflix Open Connect content delivery that saves ISPs so much bandwidth. How does it work? Why would Netflix offer to supply and install it for "free"? What do they get out of it? ... Is there a catch?

This is a simple private peering agreement where Netflix pays for the private circuits and in return gets transit for free (can send and receive traffic to the ISP's customers over these circuits for free). SK Broadband is a direct competitor of Netflix so it makes no sense for them to go for it. They would like Netflix to pay for the transit which despite every mentioning of it being heavily downvoted here is a standard negotiation item among service providers.

For example, the big 3 cloud providers not only charge for server resources (e.g. VM instances) but also for transiting traffic across their infrastructure to the Internet (and Lord, they charge a lot), effectively acting as ISPs. And if you want the private peering, easy, just pay us more. And people pay without asking!

In this case SK Broadband have a very valuable asset (8+ mln connected households) that Netflix want access to for free. SK Broadband want to monetize this asset and being direct competitors they are not trying to make friends with Netflix too.

Is this is as simple as the contention ratio in SK?

Ie. SKT previously worked with a 20:1 contention ratio. Then along came Netflix and Squid games, and suddenly they're getting complaints because it's not enough bandwidth, and so they've now having to drop that to 10:1 or 5:1 or whatever, put in faster pipes to exchanges and the like, and so it's costing them quite a lot of real money.

The Open Connect box wouldn't solve that, as the issue isn't between Netflix and SK, it's SKT to it's subscribers. And yes, although that's SKT's problem, if it's anything like the UK market and the race to the bottom, the subscribers are paying pennies.

That's conjecture on my part, and I could be very wrong. Equally I don't really know who's to blame nor how to solve it. Just an idea as to what might actually be the problem.

I think you might be exactly right. And an obvious solution would be to install FTTP to get rid of the contention issue. Or perhaps someone should set up a competitor to SKT doing exactly that. Much like Toob here in the UK.

The old BT DSL infrastructure is creaking, and just needs to be replaced.

I strongly suspect that's exactly the problem. An ISP comes up with a model that's viable based on usage patterns they are seeing at some point in time - but they actually advertise it as "unlimited", wrongly assuming that usage patterns are immutable, or (more likely) that by the time they have changed, the people responsible for the marketing will have already been paid.

Some time later, new services come along, and the usage patterns shift. The original model is no longer viable, but it has already been sold as "unlimited". Hilarity ensues.