For both wired and wireless, it's time for legislation to address exactly what it and isn't permitted with regard to advertised speeds, throttling, and total usage -- both with respect to users, and with respect to peering.
It's not just an issue of policy, but one of transparency too.
It's a fact of life that more people will want to consume data than a network can support. So where are we going to make the cut? Diminish everyone's usage proportionally? Provide a guaranteed minimum? A minimum per minute, per hour, or per day? Prioritize traffic based on type? Implement caps? Allow people to pay for more speed, higher caps, or both?
Some of these will be better or worse for different users. But since people often have little or no choice over which carrier or ISP they use and switching can be expensive, this is exactly where it makes sense for policy to be made democratically, as with utilities. And publicized.
When I pay for a carrier, I want to know what I should be able to count on, and what I can't count on.
When I pay for a carrier, I want to know what I should be able to count on, and what I can't count on.
I can't up vote this enough. I think false advertising really needs to be clamped down on across the board, but when it's something I understand the nuances of it really gets to me.
Why do companies find it so difficult to just say what it is they are selling in clear concise language, without being purposely misleading, or hiding things in fine print?
What false advertising? The ad clearly reads "up to 50mbps" and the 1.1mbps the copper in your neighbourhood allows is indeed positively in the 0-50 range!
Damn I hate consumer ISPs, with a few small exceptions.
We cover false and misleading advertising with the same laws here in Australia.
ISPs are required to give estimated primetime speeds you can expect for a given plan.
E.g. for my 100mbs plan, they tell me that in my area, I can expect 82mbs during evening peak.
We also have some ISPs who are being transparent about over provisioning and even go as far as stopping selling into an area until they acquire more bandwidth into a node.
More on topic, this is being caused by companies milking their old networks unwilling to upgrade.
I suspect it is due to execs believing their underlings when they're told their network upgrades were supposed to alleviate capacity problems at a great cost.
Fact is, usage is evolving, go to somewhere like south Korea and YouTube streaming 4k on your phone during your commute is the norm. People have the bandwidth there, and they use it.
Here, people don't have the bandwidth, so they don't, but they would.
ISPs don't seem to realise this and think it's a select few ruining the service for the whole with their pesky streaming. When really it's just changing customer needs and reluctance to serve.
> "Why do companies find it so difficult to just say what it is they are selling in clear concise language, without being purposely misleading, or hiding things in fine print?"
because it's more profitable?
it's a pretty direct line between the pressure to produce profits and lying to people to get it, particularly as it gets harder to produce profits through honest means.
remember that capitalism entices participants with profits to spur innovation, not to indulge the profit-taking itself, so any temporarily profitable enterprise should attract interest and drive down profits as competition enters the industry. so professional managers (e.g., executives) sometimes (often?) choose to lie & cheat to lengthen the period of profit-taking.
it's inexplicable that we accept that as normal, as if holding people & companies accountable for their words would stifle commerce and descend us into a totalitarian dystopia. instead, such accountability would only mean that getting filthy rich would be harder, while attaining comfortable wealth would be unchanged, if not more attainable.
You do realize that AWS SLA is about as much BS as the "unlimited" cell data plans we're discussing here, right?
All it says is that if they fail to deliver service, with a bunch of exclusions that make it very unlikely they will ever do so, then you get to not pay for that service that you didn't get.
That is very much more real than what you get from an ISP. If an ISP doesn't meet the advertised bandwidth, at best you get an apology. More likely you get nothing. You definitely aren't getting a refund for the service you didn't get.
> with a bunch of exclusions that make it very unlikely they will ever do so
I think the meat of their argument was that they have weaseled it in such a way that you're not getting what you expect unless you read all the fine print. Then it turns out you get basically nothing. From that point it's not a stretch to say the situation has not improved when compared to consumers: They also get nothing.
You can pay for a fiber line with an SLA, etc. They cost hundreds to thousands of dollars a month, even in downtown areas with several business fiber providers to choose from.
> When I pay for a carrier, I want to know what I should be able to count on, and what I can't count on.
You can absolutely count on any wireless transmission happening on a shared medium. Inconveniencing one high volume user to keep up service quality for dozens of low volume user is a perfectly rational business decision. As for transparency: well, good regulation is not easy, particularly when support for trying is not very strong.
It's a fact of life that more people will want to consume data than a network can support.
Is this actually true or is it just a matter of the isp not building the needed infrastructure required? Because if it is just an infrastructure issue then I would question why they choose to not expand to a better service. Is it because they can get away with providing crappy service as many people only have a single choice of what isp is available in their area? I highly agree with your final statement and it’s true people just want to know what they can count on and not play the guessing game.
An ISP could build the required infrastructure to support everyone's peak desired simultaneous usage... but then they'd probably have to quadruple (or worse) what people are paying for it just to break even, which the market wouldn't support, so they'd go out of business.
The same way restaurants could invest in enough space to handle all demand for Friday nights and Valentine's Day... but would go out of business because they couldn't keep it full enough to pay the rent the rest of the week/year.
Even if that number wasn’t fictional, even by its own narrative it applies to money supposedly paid to telephone companies for fiber deployment. The gist of the argument is that by promising to build fiber, telephone companies got regulators to let them increase phone charges and depreciate phone infrastructure faster than otherwise would have been the case. That’s where the $400 billion comes from.
These arguments are all based on the peculiarly of his telephone service was regulated before 1996 (and in the authors view the way it should have continued to be regulated—even though all of Western Europe went away from this sort of rate regulation around the same time the US did). That has nothing to do with cellular, which was always regulated under a totally different regime, and was built more than a decade after these events.
Basic laws of supply and demand state that when you charge a fixed price and permit unlimited usage at that price, usage will approach infinity (that is, when the marginal price is zero). Everyone with a fixed price, unlimited usage plan has no incentive at all to restrain use. The wireless carriers cannot provide infinite service (the bandwidth simply doesn't exist).
Take any public space. Now imagine it at it's peak density, when it is as crowded as it gets, maybe due to a once-a-decade event. Now imagine installing cells in sufficient density to avoid any oversubscription on that day. This density will never happen.
>Diminish everyone's usage proportionally? Provide a guaranteed minimum? A minimum per minute, per hour, or per day? Prioritize traffic based on type? Implement caps? Allow people to pay for more speed, higher caps, or both?
In the case of limited bandwidth, the "fairest" way would be to enforce a moving average cap. The most economically efficient way would be some sort of dynamic price system[1].
>But since people often have little or no choice over which carrier or ISP they use and switching can be expensive
In the context of this thread (wireless carriers), isn't this moot? I get how there are local ISP monopolies for residential internet, but the whole country has good access to all the major carriers.
[1] instead of getting 5GB, you get 5,000,000 "bandwidth credits". If a cell tower is below capacity, each megabyte of transfer only costs a nominal amount of credits (200). But as capacity approaches 100%, it costs more and more (up to 10000, or more). We could even have a total free market similar to bitcoin network fees where customers can enter a bid. If your bid isn't high enough you're limited to 2g speeds, or get no data at all.
And how is this indicated to the user? At least the ride hailing apps displayed their "surge pricing" before you used the service. As I'm traveling through town and being handed off to different cell towers, am I going to see a pop-up saying the current rate on this tower is XXXX amount and allow me to accept or not? The flexible invisible cost is just not a good thing for end users
>As I'm traveling through town and being handed off to different cell towers, am I going to see a pop-up saying the current rate on this tower is XXXX amount
Is there a reason why this can't be done? Have a carrier app that you can use to check the current rate, and that sends push notifications when the rate exceeds a rate that you've set. You can even do this through SMS so it doesn't take up any bandwidth.
>and allow me to accept or not?
Allow the user to set a maximum rate. If the current rate exceeds that rate, cut off data access. Allow the user to reactivate data or change the maximum through the app or through SMS.
>The flexible invisible cost is just not a good thing for end users
I never said it was the best approach, just the most efficient approach from a microeconomics point of view.
True, but only if they hold an monopoly. If there's a competitor, customers would switch to the cheapest option and they'd lose all their bushiness. If they're a monopoly, well... you're already hosed because they could also infinitely raise prices and there's nothing you can do.
> but the whole country has good access to all the major carriers
That's why I said switching can be expensive. If you discover bad internet speeds a month after you signed a standard 2-year contract with a carrier, what are you going to do?
Even if you pay that, you may need to pay for a new phone because of incompatible networks.
Not to mention the PITA of transferring your number.
(Similarly even if you have multiple ISP's, you may need to pay $$ for installation, a new modem, etc.)
This is your ISP being sneaky, but probably not in the way you think. This is what Netflix has been complaining about for years.
Your ISP is purposely providing a too-small link to your video provider. So when a lot of people are using that provider, it gets slow. When you switch to another provider, or a VPN, you get to use a fast connection to that other VPN that isn't overloaded, and then a fast connection back to your video provider.
This way your ISP can claim they are not throttling traffic because they aren't inspecting the packets for traffic shaping, they are just using physically poor connectivity to certain providers for traffic shaping.
Not to mention, they can now go and ask that provider to pay them for the right to 'upgrade' their connection, to enable their customers to get what they are paying the ISP for.
Yup, and in Comcast's case, when Netflix finally paid Comcast's ransom, the problem was suddenly fixed a few minutes later, because Comcast literally just had to reconfigure the router and plug in a second cable.
In those days it was typical for the costs of peering to be borne by the sender, relative to the traffic ratio. This changed when Netflix and Youtube decided to use their PR and lobbying strength to bring the negotiation to the court of public opinion - and normal people started referring to typical peering cost-sharing agreements using emotionally loaded words like "ransom"
Okay, let's go the other route then. BackBlaze, CrashPlan. These are companies where Comcast/other ISP customers are sending THEM huge amounts of data and getting nearly nothing in return.
So these ISPs should be bearing the costs of peering to backup providers, correct, by your logic? Or would that be the backup providers holding ISPs to "ransom"?
Apropos of anything else, ISPs (generally) provide massively asymmetric pipes. You can't provide an asymmetric pipe and then complain that data use on that pipe is, well, asymmetric.
Certainly your ingress/egress ratio affects what kind of peering deals you can get. Peering can be something of a wild west. If you can charge the other party, you do. Otherwise you don't.
The ratio requirements are a red herring. If your ISP says you get a certain bandwidth, it is on them to provide the appropriate pipes to get to where their customers want to go. How can the ISP on the other side of Comcast be responsible for maintaining ratios if Comcast's customers are the ones that are requesting the data?
>appropriate pipes to get to where their customers want to go.
This idea is nonsensical. What if I want to go to somewhere that is three times removed from my ISP. Comcast has zero capability to force some South Korean ISP that is providing services to some website to increase their bandwidth with their upstream who is peered with whoever owns the undersea cable that is peered with the company that peers with Comcast.
The alternative is that Netflix asks for 40 Gbps to Comcast and suddenly Comcast is obligated to provide 40 Gbps to every one of its customers.
What we've always gotten with an ISP is a given link rate to our ISP. Beyond that it's always been a crap shoot.
If everyone up to the network edge is willing to pay for the equipment to get the packets there, and your customers are asking for the data, then yes, you have an obligation to deliver what you promised to your customer.
So if Netflix were sitting on the end of a lowly T1, Comcast should provide Netflix with a more appropriate connection that can fulfill the demands of Comcast customers?
At the time, Netflix didn't have a deal with Comcast. There was a company between the two (Cogent?) that wasn't purchasing enough bandwidth for what they were selling. If you were on Comcast and watched Netflix via an Apple TV, it worked fine because for some reason Apple TV streams were sent via Level 3.
The end result was Netflix doing a deal to connect directly with Comcast, cutting out the middleman (IIRC).
That's not really accurate either. Netflix's ISP didn't have the bandwidth available. That was the whole dispute. Their ISP wanted a free peering agreement and Comcast just refused.
Netflix bought bandwidth from more than one company. That's why if you used a VPN or a different type of device for streaming, it would work. Comcast wasn't throttling Netflix in general, they had an oversubscribed link with Cogent.
They had sufficient bandwidth. Netflix paid their ISP for sufficient bandwidth. Comcast just refused to connect to it, to the detriment of Comcast's customers.
Cogent (Netflix's ISP) and Comcast had a peering agreement which spelled out costs and ratios; Cogent sold Netflix bandwidth (probably very cheaply) that they could not provide without breaching their peering agreement. Cogent was unwilling to fulfill the financial terms of their peering agreement in order to provide the product that they had sold to Netflix. Somehow, people think this is Comcast's fault?
Netflix is a very popular company and Comcast is a deeply unpopular company. Consequently people take Netflix's statements at face value while assuming that Comcast always lies, even in situations where the facts suggest otherwise.
But who's fault is any of this? I say you reap what you sow. Comcast has only themselves to blame when consumers are skeptical of them.
You can also turn that statement around. Customers have paid Comcast to access the internet (Netflix) and Comcast didn't want to upgrade his peering with backbone providers forcing Netflix to pay them interconnection.
In this case Comcast is getting money from it's customers and then Netflix has to pay extra for the traffic Comcast's customers requested and paid for.
...the costs of peering to be borne by the sender, relative to the traffic ratio.
This "rule" is too simple to game, to have ever been a functioning rule. Netflix could have just patched their app to constantly upload lots of random data. This would have killed ISPs who often expect mostly-down traffic profiles, but it would have evened out "the ratio" so it theoretically would have won the peering argument. Of course, Netflix never did this, which proves that this so-called rule never really governed anything.
One of my biggest factors for choosing an ISP and contract is how much bandwidth streaming a high-res movie is going to take.
I had to chose Comcast when I moved because the alternative could only handle HD streaming but not 4K. And then I'm paying half again as much to be able to watch a pretty movie while others are using the internet, or for several of us to watch shows on our own.
Peering ratio requirements are kind of a scapegoat. Especially in the case of an 'eyeball' network like Comcast. I have 250Mb down, and 10mb up. (I would love 50Mb up, I push a ton of docker images daily). Their circuits they sell their clients will only re-enforce those ratios.. (then again, if netflix was to partner with a bunch of backup providers, like Mozy, crashplan, backblaze, etc, to carry their backup traffic, that could get VERY interesting)
Yeah, but these consumer ISPs self-inflicted the ratio imbalance. They literally don't let their customers upload at the same speed they download at; of course their peering ratios are going to be imbalanced.
I checked several of the largest ASNs and several smaller NYC-area ISPs on peeringdb and they are all "No" to the ratio requirement. Google doesn't care. Amazon doesn't care. Cloudflare doesn't care. There's half the Internet.
Demanding money to peer with Netflix is just trying to charge customers twice because they can. It's not normal, conventional, or good for the Internet. Netflix might be able to pay, but will your startup be able to pay? Probably not. And that's not the future we want.
And if Netflix and Youtube were being charged the actual charge for a tech to walk down and plug in another cable or pop together another interconnect, that would be fine.
Instead companies like Comcast are seeking to make profit from this.
All of the companies involved are seeking to make a profit. For example, Netflix and Youtube are seeking to lower their cost of bandwidth, thus increasing their profits.
Same phenomenon for me in southern Ontario. My ISP (Start.ca) is using Rogers' infrastructure. Evenings equal poor streaming video performance on all sites despite an otherwise-perfect connection. If I connect to a VPN, video performance mysteriously improves instantly.
If I were of a suspicious mind, I'd wonder if Rogers is pulling some backend shenanigans to discourage cord-cutting.
There is an actual problem here, but wireless carriers use the actual problem of how to balance traffic as a fake cover for charging more.
However, assuming we bypass the garbage tiering, there is a legitimate reason to do this. Page-loads are simple and fast. Every one loading pages is not much of an issue, as they typically load at different times for a brief moment and then are done. This means for a rated connection speed, you don't need as much actual capacity.
Video streaming (or other downloads) are constant. To stream video at rated speeds, you must build to provide most users that speed at the same time. Problem two here is that in cases of high traffic, video streaming is greedy. If it's not going at full resolution, it will suck up all available bandwidth it is given. If more is given, it will raise quality (by default settings, at least).
A sensible solution to this is to stream video and audio over UDP and have the router just drop UDP when TCP needs bandwidth (to a point). The point is, requirements are different. People sending e-mails shouldn't have to wait a minute to send because you really want to stream your netflix every where you go and in 4k.
Which reminds me, I thought mobile connections were for simple stuff. Light browsing, email, what-have-you. Even with "unlimited" plans, you're throttled after twenty-odd gigabytes of data per month. Why can't people just download netflix in advance? Failing that, I'm not sure why people expect to use a wireless internet connection in the same way as a wired one.
if I were legislating this I'd mandate transparency before I mandated neutrality -- maybe consumers would be willing to pay up or switch to a provider that provided neutrality for streaming.
one way for government to provide usable markets is to mandate good information without setting standards.
NFLX is over neutrality for home internet (not sure about mobile), they're in every server cage
If you have a monopoly or duopoly, transparency doesn't really help. Your only service provider transparently screwing you doesn't give you an option to switch ...
It surprises me that most people in the US wouldn't have at least one cable and one DSL option. After all, cable TV is available most places, and everyone has a phone line (or is at least wired for one).
Or is the issue rural locations which were never wired for cable because they all just got satellite dishes?
I live in a large US city and have one cable provider as my only viable high speed option. The dsl provider which I would have preferred hasn't upgraded the last 50 meters of lines going to my house, and said they had no plans to. The top speed they could offer me was 3Mbps.
EDIT: In comparison, the same dsl provider on streets that have had their lines upgraded offers 100Mbps plans for about $30 less per month than the cable provider. So some homes actually do enjoy a little healthy competition, but it hasn't been enough to cause any changes in the cable plan pricing yet :(
You'd think there might be a business opportunity in this kind of situation for an ultra low cost, short range WISP. Like, if you could identify a bunch of scenarios like this where last mile issues have left behind neighbourhoods where there might be a few dozen potential customers (identify using speedtest.net data or something?), it would be possible to plug in a box on a utility pole at the edge of the existing coverage and then supply the participating households with modems. I don't know what the cost of that kind of infrastructure is or how much of an issue it would be to plug into what's already there, but hopefully it's not terrible assuming the existing provider would agree to let you in.
I guess the main risk is that doing this might wake up the provider who would then finish their rollout to get those customers back, and undercut you on price. So it would need to be quick and easy to grab your box and move it to some other neighbourhood in that scenario.
...plug in a box on a utility pole at the edge of the existing coverage and then supply the participating households with modems. I don't know what the cost of that kind of infrastructure is...
It's very expensive.
...or how much of an issue it would be to plug into what's already there, but hopefully it's not terrible assuming the existing provider would agree to let you in.
They won't let you in.
I guess the main risk is that doing this might wake up the provider who would then finish their rollout to get those customers back, and undercut you on price.
And they also do that. National ISPs can afford to lose money in one city in the short term but you can't.
The pole rentals are extremely expensive and the existing players aren’t required to play with you.
A few players like this tried to use streetlights for this sort of thing with WiFi. Hanging a WiFi point on a pole is outrageous unless the city owns the pole and plays ball.
I live 3 miles from downtown Austin, which has numerous internet options (Google Fiber, Spectrum, AT&T, and Grande), but the options in my apartment complex are:
* Spectrum (up to 400mbps)
* AT&T (5mbps dsl)
I technically do have some choice, but given that one of these is close to 100x faster than the other, and the pricing is fairly similar...
I think wireless isps and 5g are our best chance at disrupting this.
I used to live 8 miles from Madison, WI (technically, one exit south of their beltline interchange on I-90). My ONLY two choices were 2mb WISP for $65/month, or 4g hotspot, and all of them were VERY expensive for over 15GB a month. (I used about 20GB per month on the 2mb wisp because I had little kids who loved netflix cartoons, faster speed would also cause higher quality video)
There was NO cable or dsl. ATT always said DSL was available in our area, and that they would schedule the install, then 3 days later, the local guy would call, and say "oh, hell no, lol" and cancel it
I don't know if this was viable at the time, but nowadays I expect you could use an OpenWRT router to do per-domain/mac traffic shaping on the device displaying the cartoons. That would force it down to 480p (or whatever) and reduce your overall usage.
Another option that might not have been available at the time would be taking the Netflix device to a different connection (work, the library, whatever) and then pre-downloading a bunch of the cartoons for offline viewing.
Oh gross, yeah, multi-unit buildings are a funny case where the choices are severely reduced due to factors outside your control. And it's not like a condo building where your HOA can demand better options— as a renter, you really do have no say in the matter.
I wonder why the cable option is so bad? Is it just because it's a single coax entering the building shared by all residents? Where I am (Kitchener ON), I can get 200mbps cable.
I have family living in a motor home, and they deal with something similar at RV resorts; they're stuck with the wifi available to them there, and little ability to use an alternative (short of cellular).
Even that spectrum connection is garbage because the upload will only be 2 or 3 mbps, barely enough for one FaceTime call. Anything short of symmetric fiber is a sham in this day and age.
It's probably 20mbps shared between the whole neighborhood with high latency.
At this point, we should have symmetric fiber with at sub 50ms latency and 100mbps down and up everywhere like we work to have water, sewage, electricity, and gas. Right now, all of our money goes towards maintaining an antiquated system for providing commercial laden garbage media.
A 20mbps shared limit for an entire neighborhood for a major cableco might have been the case 20 years ago when cable internet deployment was in its infancy.
While I agree that symmetric fiber would be better, my speeds are always consistently close to what I'm paying for and I'm one of those types of people that:
1) utilizes my upstream because of things like DR backups
2) treats cabelco with suspicion and thus records speeds, latency, packet loss stats continously
3) will pick up the phone and complain when I see problems until they are resolved to my satisfaction
In downtown Portland, OR my options are: Comcast with a monthly data cap and irregular throttling, Centurylink with a monthly data cap and irregular throttling at about the same price as Comcast, or a variety of minor providers with even stricter monthly data caps and/or other limitations inherent in service (eg. satellite internet).
...rural locations which were never wired for cable because they all just got satellite dishes
This has reversed the actual cause and effect. Zero rural customers are served by cable. Some rural customers have satellite dishes.
There are numerous reasons why cablecos don't serve rural areas. The obvious reason is the expense that would be required to run a mile of copper for each customer. Less obvious reasons include the fact that if there is no municipal apparatus to target for corruption, the cableco business model sort of falls apart...
You don't even have to visit a rural area to see this. There are lots of satellite customers in cities and that fact hasn't chased off the cablecos.
Maybe 10-15 years ago. Today, the traditional telcos are just letting the cooper service rot.
In my city the copper service had a significant 3 week outage that when a CO switch failed. They would rather attrit the CWA people and pay a fine than maintain it. Verizon’s position is to wait for direct subsidy or 5G in urban areas.
Not available anywhere I've lived (Bay area, Chicago, SF). In fact I've only heard about it from Wikipedia. Never heard of anyone who knows anyone who claims to have even seen this in person.
I live in a smallish suburb (~5,000) of a small city (~150,000). I have access to Comcast. That's it. AT&T won't even sell us a DSL line, just satellite. There are no WISPs within range. LTE is far too expensive for our usage patterns.
for wired internet I suspect this is right in some places -- I read a thing about fiber providers slicing up cities by block or zip code
but for wireless most providers have dense coverage maps. I use ting which operates on sprint or t-mobile I think so there's even more variety if you include resellers.
I’m reminded of a comment I read a few days ago from someone that has their own small isp that said bandwidth was basically an inconsequential cost for them. Throttling like this is penny pinching of the highest order. How do we get the benefits of competition if every company is doing it?
Cable companies and wireless companies want to keep selling you your data multiple times. Internet fast enough for video eats in to selling you video. Wireless internet that can use WhatsApp for mms means you won't pay extra for more texts.
I really, really hate to defend mobile carriers, but...
Congestion is a real thing. Mobile subscribers are now using significantly more data due to video streaming, and wireless carriers aren't expanding their capacity fast enough. 5G won't help for most of CONUS. Millimeter wave 5G, which will help, will only be deployed in urban areas. Mid-band and low-band 5G will only increase spectrum efficiency around 20%.
For now, carriers are throttling videos and introducing various deprioritization options. All the major carriers only offer plans with forced traffic deprioritization. For some plans, all traffic is deprioritized. For others, deprioritization happens after a threshold.
There isn't enough bandwidth in the US to support 240 million subscribers streaming Netflix 24/7.
- I feel like there should be some oversight or required transparency about how and when they're applying throttling. Can they throttle customers in under served markets all the time to save money?
- They're currently throttling sites not protocols or users. Why? Did some sites pay them off? If they need to throttle they should be applying that process equally among sites. "But AT&T didn't slow Amazon.com's Prime Video at all"
- I would be curious to see where bandwidth problems play out in terms of "carrier could build more towers, but hasn't" and "we've reached the actual peak of what's possible with current technology.
> I really, really hate to defend mobile carriers, but...
For the most part, carriers sell unlimited, then throttle heavy users. While it's not honest, they're also not upselling those users on super unlimited plans. This makes me think carriers genuinely don't have the capacity for all this video. Without an upsell path, if they had the bandwidth, there'd by no incentive to throttle.
> Not at all. TCP already does its own rate limiting without external throttling.
Individually, cars brake on the freeway to avoid hitting the car in front of them.
Collectively on a busy freeway, that sets off a chain reaction that results in the familiar rush hour crawl.
Minnesota did a study in 2001 on the effect of ramp meters, and disabling the meters (ie, removing throttling) resulted in statistically significant increase in overall freeway travel times.[0]
Statistically multiplexed shared networks like mobile wireless face similar issues. For a single TCP session the only metrics that can be divined by the endpoints are round-trip time and loss. As the shared network reaches capacity, larger numbers of TCP connections all back off around the same time, but large flows are more aggressive at ramping up than smaller flow (email/instant messaging, etc) and can result in an effective breakdown of network usability. A network control that has visibility to multiple flows and awareness of capacity of the system can influence overall performance much more effectively.
Ideally mobile providers would be trying to shoot for maintaining uniform latency per flow, similar to what queuing strategies like CoDel[1] achieve, but that's likely beyond the CPU and buffering capabilities of their existing hardware. Lacking the perfect solution, it's human nature to move on to the next approach: managing the biggest problem. Video is usually easy to identify, and so it wins the "able to be managed" prize.
This is somewhat true, but there's a lot of work done in the data link layer to provide a robust connection. TCP falls over with packet loss of less that 1%[1], so it's not really good at that, anyway.
Anyone here remember how Google/Netflix sold us out when they made deals with AT&T and the other ISPs and suddenly became very quiet about net neutrality?
Many warned then that they it's foolish to think that they can get ahead of competitors this way, but they didn't listen. They had this coming.
I don’t necessarily want throttling at the ISP level but I do want rules on what is “reasonable” for web sites on wireless. Data is expensive.
A big one: I often do what should be simple things data-wise, like tapping a link to “view article”, and am punished. Is it my fault that the obnoxious target site decided on its own to download and auto-play a huge video with sound in a floating window, none of which was requested or required!?
I have the right to squash that crap to save my data plan and battery (and sanity). That has meant installing blockers, and it doesn’t look like I’ll ever be able to stop doing so.
If these streaming providers would actually provide consumers with effective tooling to determine if throttling is taking place, this would get a lot more traction. I fought with CenturyLink for 18 months collecting stats on overall network performance vs. throughput to YouTube because it was clear something was going on. It was a giant pain in the ass trying to come up with figures for YouTube.
Check out https://fast.com from Netflix. It's a speed test based off of Netflix's servers, so it will tell you your bandwidth to Netflix at any given time.
Then complain to your ISP if that doesn't match your advertised speed.
When the advertised speed is "up to XX Mbps" and you occasionally do get those speeds, then what is the proper complaint? They are providing "up to" the speed purchased. It works very well in the provider's favor.
I guess the best you can do it test it multiple times and then complain that you've never seen over XX Mpbs and then ask when exactly you'll get "up to" the advertised speed.
FWIW, even if you don't complain, it will help Netflix gather data that they can use to complain on your behalf.
There is no way to complain, but if we had some open, trusted services similar to speedtest.net to publish the effective speed it would allow people to make informed choices when selecting an ISP.
Just ran it and I'm getting 44 Mbps. AFAIK, that's plenty of bandwidth from Netflix's perspective, so would complaining to my ISP make any sense? Should I be getting 300 Mbps with Netflix?
If you pay for 300 you should get close the that. The fact that you don’t means you’re being throttled. If you don’t notice a difference it doesn’t really matter, but that wouldn’t be enough to stream multiple TVs at 4K for example, even though you pay for a service that should support that.
This is where net neutrality fails. I'm all for not letting vertically integrated carriers give their own services an unfair advantage, but if throttling a high-bandwidth user to improve the service of nearby users? I'm more ok with that.
I don't see an issue with throttling video at all - the idea that I can bang a 4K stream down on LTE, which might be limited to 100mbit from the tower for everyone, means that's necessarily going to happen.
That said, I use secure encryption to prevent the ISP from viewing my browsing habits. Throttling VPN use carte-blanche because you don't know what's in there would be overkill.
It seems to me that it's best to just throttle based on allowing burst usage and treating all bulk downloading including streaming video identically.
Here's a thought - just mandate that all carriers should have an API that returns true congestion rate and throttling rate of users at a given time. This way we can keep everyone honest and transparent.
That way - If i'm unable to watch my Netflix show in HD without bufferring across LTE, Netflix app can call an API to see if the throttle is limiting them.
I see a number of people who use YouTube just to listen to music on their phone. If you have an unlimited plan it's cheaper than paying for Spotify. So instead of a cheap 128kbps stream they are sending full motion video for someone who isn't watching anyway. I am curious what portion of streaming this represents.
If companies were better about being up-front, I would have no issue. I usually can't tell when my phone's streaming resolution drops to 480p, so I really don't care. It's being sneaky about it, on page 13 in small text, that bothers me.
Wireless is a medium that is severely limited because of the laws of physics. I 100% support QoS.
And by QoS I don't mean limiting video even if there's available bandwidth, I mean limiting it if other people are trying to use their connections reasonably.
In part, because most video sites will adapt the stream to the achievable bandwidth.
In part, because if the throttle all bulk data transfers over some minimum size, all the time, most people are only going to notice on video streaming, since videos are the most frequent bulk transfers.
> In part, because most video sites will adapt the stream to the achievable bandwidth.
Why is that a reason to force them to use less than their fair share of available bandwidth?
> In part, because if the throttle all bulk data transfers over some minimum size, all the time, most people are only going to notice on video streaming, since videos are the most frequent bulk transfers.
How is that a reason for anything, and what is it actually supposed to be a reason for?
>> In part, because if the throttle all bulk data transfers over some minimum size, all the time, most people are only going to notice on video streaming, since videos are the most frequent bulk transfers.
>How is that a reason for anything, and what is it actually supposed to be a reason for?
I was unclear. I'm suggesting that carriers are in fact, throttling all bulk data, but reporting is focused on video. Allow a sizable burst of data at the beginning of a connection and then start rate limiting, and most people won't notice except for on videos.
This was probably the simplest, least noticable, least costly intervention to reduce peak congestion and provide a consistent experience across the carrier network.
Is it fair? Maybe, if applied to all traffic
Is it clear and transparent? No, clearly not.
Does it allow for the network to be used to capacity? No, not unless the limits were modulated based on current use, which doesn't seem to be the case.
Is it good for customers? Unclear -- to the extent you can watch good enough videos using less of your data quota, that might be a good thing; to the extent that you spend more time downloading updates, that's probably not good, but would need analysis of battery impact traded off with better push latency if the radio is kept on for other reasons. Also, to the extent that this throttling provides better availability in congested area, that's a plus for users.
> I was unclear. I'm suggesting that carriers are in fact, throttling all bulk data, but reporting is focused on video.
Seems unlikely to me, but who knows ...
> Also, to the extent that this throttling provides better availability in congested area, that's a plus for users.
If you are being throttled, that's obviously not increasing availability in any meaningful sense, so, no, it's not. What would be a plus for users would be adding capacity. Not performing the contractually agreed service is the exact opposite of "increasing availability".
Overall, the problem with all of this is that it borders on fraud. There would be no problem with selling low-bandwidth plans, if that is the only thing you can offer at economical prices. Or offering the user the choice to throttle certain services so as to conserve bandwidth. The problem is that carriers kinda-sorta sell a service and then don't do what they kinda-sorta promised. And a big part of the problem is that everyone frames this as "trying to help the customer by limiting abusers", or something similar, when really, the only abuser is the carrier selling a service they don't intend to perform, because they would have to spend money to do that.
The important part is not whether some traffic is throttled, the important part is who makes the decision as to what gets throttled. If I buy 1 Mb/s mobile IP access, and then the carrier throttles my connection to 1 Mb/s, that's my decision. If I buy "LTE full-speed IP access", and then the carrier throttles my videos to 1 Mb/s with no option to opt out, it is not.
> If you are being throttled, that's obviously not increasing availability in any meaningful sense, so, no, it's not. What would be a plus for users would be adding capacity. Not performing the contractually agreed service is the exact opposite of "increasing availability".
Adding capacity is easy to say, but can be hard to do. Where I live, the city council basically won't permit new towers; whatever licensed spectrum on the towers we have is the capacity.
Trying to frame this as a contractual issue is a losing battle. Nowhere ever did someone contractually promise you a meaningful speed to an unknown destination at a consumer approachable price. Certainly, it's not transparent, and that's reason to be upset.
In a world (like ours) with limited bandwidth, you sometimes want to limit flows in a way to produce the least harm. Throttling different streams has different effects on user happiness.
For instance, if you are 100 Mbits over capacity, you could either:
(a) Throttle 100 users Netflix streams, which drop from 1080p to 720p
(b) Throttle email, it backs up causing hour-long email delays.
That is actually the first mistake in your argument in this context. The bandwidth limits in mobile networks are first and foremost an effect of decisions by the network operators, not an inherent property of the world.
> you sometimes want to limit flows in a way to produce the least harm. Throttling different streams has different effects on user happiness.
Well, maybe. But I would argue that that is both (a) prone to erroneous valuation, if only due to conflicts of interest, and (b) almost always only the case to any significant degree in very artificial scenarios, not in the real world.
> For instance, if you are 100 Mbits over capacity, you could either:
> (a) Throttle 100 users Netflix streams, which drop from 1080p to 720p
> (b) Throttle email, it backs up causing hour-long email delays.
> One is obviously less harmful.
Primarily, that's a nonsensical scenario. For one, you are never "over capacity". That's what capacity means: The maximum speed at which you can transfer data. You never transfer data faster than you can. You are only "over capacity" in the sense that the link is congested, but that isn't a data rate that you are "over the limit", it's just that the queue starts to grow.
But also, your scenario implies that throttling email more than its fair share somehow would cause "hour-long email delays". Now, I dunno, what would that look like? 100 users using netflix, ~ 10 Mb/s each, for a total of 1000 Mb/s, and one user sending emails at 1000 Mb/s, which would then lead to the email sender being throttled to 950 Mb/s under fair sharing, which ... still would not lead to anything remotely like "hour-long email delays", despite being a totally contrived scenario?
It looks more like you just made up one bad thing and one not so bad thing, and then noted that the worse thing is worse. Which is fine. But it is completely irrelevant to the discussion if you do not demonstrate how those two options are actually options in the same situation, and how they, respectively, compare to fair sharing.
It's an inherent property of the world that bandwidth is expensive, so it's always going to be limited.
There are two types of protocols: those that adjust the quantity of data sent depending on conditions, and those that don't. Streaming video does -- if throughput drops, it changes video encoding to send fewer bytes. Most other protocols don't -- they just get slow but they still need to send all the bytes eventually. There's a solid argument for throttling the first kind when congestion hits.
> It's an inherent property of the world that bandwidth is expensive, so it's always going to be limited.
Which is about as vague as it can get.
To make things more concrete: What would a gigabyte of mobile traffic cost if network operators were to optimize for low per-GB cost?
> There are two types of protocols: those that adjust the quantity of data sent depending on conditions, and those that don't. Streaming video does -- if throughput drops, it changes video encoding to send fewer bytes. Most other protocols don't -- they just get slow but they still need to send all the bytes eventually. There's a solid argument for throttling the first kind when congestion hits.
Except: There really is not, especially not between customers, which is what this discussion is about.
It is not some random accident that video streaming happens to be the thing
that adapts to available bandwidth: That is precisely because it eats so much bandwidth compared to everything else. Which also means that throttling it in favor of other applications between different customers is actually not really useful. If you have one customer that watches a video stream and one (comparable, end-user) customer sending emails, then the email user will pretty much always have lower bandwidth requirements than their fair share anyway.
Also, it simply is not up to you to decide what a customer should value. If you sell a customer a contract that promises a certain bandwidth that is sufficient for high quality video streaming, then it is your job to provide that bandwidth at any time, except for rare occasions when/where unexpected demand hits you. It is not up to you to decide that a low-quality stream ought to be good enough, or that someone else's emails are more important than their high-quality video stream, if they are paying you for the same level of service. If the customer thought that that would be good enough, they would have bought a lower-bandwidth plan. If you don't offer a lower-bandwidth plan, then that is your fault.
Congestion should never be a standard condition in your network, if it is, what you are doing is essentially fraud by selling people a service that you don't intend to perform. The fact that it might be technically impossible to perform the service doesn't change that, it's still your fault if you know that you can't perform the service, but you enter into contracts anyway. Congestion should be an exceptional condition, and as such should never be so massive as to have any advantage from using application-dependent throttling between customers.
I'm not sure what your question really is?! Is it whether I would want to buy an internet connection that's too slow for watching most video? Or whether I would want an internet connection that is too slow for watching most video to be offered on the market for people to buy? Or ... what?
Because in case of congestion the user watching video is abusing the system. It's like going to a congested public road with a convoy of lorries just because you can. Really, don't be a dick.
Because in case of congestion (let's say everybody gets allotted 50 KB/s) you can't even watch video. Dozens of users trying to watch video at 50 KB/s aren't even watching video; they're just launching a DDoS attack. I'd be very happy to throttle them to 1 KB/s so I don't have to wait 10 seconds to receive a picture through Whatsapp.
You're arguing against the laws of physics here, and you're arguing that stupid users should be able to screw everybody else. No thanks.
> Because in case of congestion (let's say everybody gets allotted 50 KB/s) you can't even watch video.
You can watch video at 500 kbps perfectly fine.
> Dozens of users trying to watch video at 50 KB/s aren't even watching video; they're just launching a DDoS attack.
So, you are saying the problem is with people who can't watch video, but do watch video anyway, despite the impossibility? Or are you saying that people who can't watch video, and therefore don't watch video, are congesting the network by not watching video? I can't really figure out what your claim even is.
> I'd be very happy to throttle them to 1 KB/s so I don't have to wait 10 seconds to receive a picture through Whatsapp.
OK, I understand that you would like to get preferential treatment. That's fine, but does not particularly convince me.
> You're arguing against the laws of physics here,
Which law of physics exactly?
> and you're arguing that stupid users should be able to screw everybody else. No thanks.
How is getting the same amount of resources as everybody else for paying the same price as everybody else "screwing everybody else"?
Really, it seems to me like you are simply repeating a completely unjustified claim over and over in different wording ... you might want to consider that repeating a claim does not make it more convincing, especially if you don't address objections.
>You're arguing against the laws of physics here and you're arguing that stupid users should be able to screw everybody else. No thanks.
i think there is a misunderstanding here.
he's not arguing that ISPs shouldn' be able to bandwidth-discriminate based on customer. He's arguing that they shouldn't be able to discriminate on the type of data. what would it matter to the other users if they were slowing down the network by html5 video, VNC, or playing a game over TCP?
I don't know a lot about networks, but i think a potential solution could be to discriminate between high-bandwidth and low-latency transmissions. For a video game or something you would want to have low latency, but you wouldn't need a lot of bandwidth. on the other hand, video streaming requires a lot of bandwidth but doesn't depend so much on latency as long as it has a big enough cache. images are probably somewhere in the middle. this would allow low-bandwidth applications to be prioritized.
> this would allow low-bandwidth applications to be prioritized.
That is a completely different kind of "prioritization" than what we are talking about here, though.
In particular, this "prioritization of low-bandwidth applications" does not in any way need to be aware of applications. All you need to do is to do fair link scheduling over all customers on a given link with some sort of token bucket per customer that allows short bursting. That way, if there is a customer that's been idle for a few minutes, say, they'll be able to transfer at, say, ten times the fair average rate for a few seconds, while other customers are slowed down proportionally, thus providing them with low latency. But that does not imply a higher average rate, nor a distinction of video vs. non-video, or anything like that. On average, every customer on a congested link would still be getting the exact same bandwidth, just as a constant low-bandwidth stream for some and as a series of short, fast bursts for others.
What tcxgy is arguing for is that if you are streaming video (or whatever other application that he doesn't like), then you should only be getting 2% of the average bandwidth compared to someone who is using an application he approves of. So, if you are constantly posting high-resolution pictures (something that he seems to approve of) at an average of 50 kB/s, that should get 50 kB/s, but if you are watching a 25 kB/s video stream of some sort, that should be throttled to 1 kB/s, because otherwise it's supposedly violating some laws of physics, and it's abusive to use half the bandwidth for video, and whatever other nonsense arguments he has come up with so far.
If the customer specifies QoS, why not? There are 6 reserved bits after HLEN in a TCP packet header, which could be used to annotate the priority of the stream - one for low-latency but also low bandwidth (=Whatsapp, messengers, gaming, VoIP), and one for bulk transfers (=high latency acceptable, high data volume).
Default would be the first so that networks may only have special treatment for apps that explicitly opt in to say "i'm ok to wait a bit" like Netflix, Youtube and friends.
On the one hand, I can see needing the ability to throttle for network traffic handling because you'll always have that one person who goes into an all you can eat, grabs 10 plates and all the chocolate pudding because they have to feel like they got one over on ... well, everyone. On the other hand, doing this 24/7? That's just not right and I think this runs afoul network neutrality legis... oh, wait.
> On the one hand, I can see needing the ability to throttle for network traffic handling because you'll always have that one person who goes into an all you can eat, grabs 10 plates and all the chocolate pudding because they have to feel like they got one over on ... well, everyone.
So, you want to allow companies to not fulfill contracts because the company has failed to fulfill contracts?
Mind you, if customers in this scenario of yours don't get to eat as much as they can eat, that has absolutely nothing to do with the other customer that happened to eat as much as they can eat, and everything with the company not being able to fulfill the contract that they entered into. It is the company's fault, and only the company's fault, if they enter into contracts with multiple people that say "N bucks for as much as you can eat", if they then don't have enough food to feed all those people.
No, that's not what I said at all. I want someone tending the eatery telling the person scooping out all of the chocolate pudding that, sure, they may have as much as they like, but let others have some now and we have more in the back that we're bringing out. You just need to wait a few moments, so have your one plate, then come back for more.
It's not just an issue of policy, but one of transparency too.
It's a fact of life that more people will want to consume data than a network can support. So where are we going to make the cut? Diminish everyone's usage proportionally? Provide a guaranteed minimum? A minimum per minute, per hour, or per day? Prioritize traffic based on type? Implement caps? Allow people to pay for more speed, higher caps, or both?
Some of these will be better or worse for different users. But since people often have little or no choice over which carrier or ISP they use and switching can be expensive, this is exactly where it makes sense for policy to be made democratically, as with utilities. And publicized.
When I pay for a carrier, I want to know what I should be able to count on, and what I can't count on.
I can't up vote this enough. I think false advertising really needs to be clamped down on across the board, but when it's something I understand the nuances of it really gets to me.
Why do companies find it so difficult to just say what it is they are selling in clear concise language, without being purposely misleading, or hiding things in fine print?
Damn I hate consumer ISPs, with a few small exceptions.
ISPs are required to give estimated primetime speeds you can expect for a given plan.
E.g. for my 100mbs plan, they tell me that in my area, I can expect 82mbs during evening peak.
We also have some ISPs who are being transparent about over provisioning and even go as far as stopping selling into an area until they acquire more bandwidth into a node.
More on topic, this is being caused by companies milking their old networks unwilling to upgrade.
I suspect it is due to execs believing their underlings when they're told their network upgrades were supposed to alleviate capacity problems at a great cost.
Fact is, usage is evolving, go to somewhere like south Korea and YouTube streaming 4k on your phone during your commute is the norm. People have the bandwidth there, and they use it.
Here, people don't have the bandwidth, so they don't, but they would.
ISPs don't seem to realise this and think it's a select few ruining the service for the whole with their pesky streaming. When really it's just changing customer needs and reluctance to serve.
Of course during the day when people actually use the network it's down to 0.5Gbit, down to 0.3Gbit during prime Netflix hours.
because it's more profitable?
it's a pretty direct line between the pressure to produce profits and lying to people to get it, particularly as it gets harder to produce profits through honest means.
remember that capitalism entices participants with profits to spur innovation, not to indulge the profit-taking itself, so any temporarily profitable enterprise should attract interest and drive down profits as competition enters the industry. so professional managers (e.g., executives) sometimes (often?) choose to lie & cheat to lengthen the period of profit-taking.
it's inexplicable that we accept that as normal, as if holding people & companies accountable for their words would stifle commerce and descend us into a totalitarian dystopia. instead, such accountability would only mean that getting filthy rich would be harder, while attaining comfortable wealth would be unchanged, if not more attainable.
All it says is that if they fail to deliver service, with a bunch of exclusions that make it very unlikely they will ever do so, then you get to not pay for that service that you didn't get.
http://aws.amazon.com/compute/sla/
I think the meat of their argument was that they have weaseled it in such a way that you're not getting what you expect unless you read all the fine print. Then it turns out you get basically nothing. From that point it's not a stretch to say the situation has not improved when compared to consumers: They also get nothing.
You can absolutely count on any wireless transmission happening on a shared medium. Inconveniencing one high volume user to keep up service quality for dozens of low volume user is a perfectly rational business decision. As for transparency: well, good regulation is not easy, particularly when support for trying is not very strong.
Is this actually true or is it just a matter of the isp not building the needed infrastructure required? Because if it is just an infrastructure issue then I would question why they choose to not expand to a better service. Is it because they can get away with providing crappy service as many people only have a single choice of what isp is available in their area? I highly agree with your final statement and it’s true people just want to know what they can count on and not play the guessing game.
An ISP could build the required infrastructure to support everyone's peak desired simultaneous usage... but then they'd probably have to quadruple (or worse) what people are paying for it just to break even, which the market wouldn't support, so they'd go out of business.
The same way restaurants could invest in enough space to handle all demand for Friday nights and Valentine's Day... but would go out of business because they couldn't keep it full enough to pay the rent the rest of the week/year.
Even if that number wasn’t fictional, even by its own narrative it applies to money supposedly paid to telephone companies for fiber deployment. The gist of the argument is that by promising to build fiber, telephone companies got regulators to let them increase phone charges and depreciate phone infrastructure faster than otherwise would have been the case. That’s where the $400 billion comes from.
These arguments are all based on the peculiarly of his telephone service was regulated before 1996 (and in the authors view the way it should have continued to be regulated—even though all of Western Europe went away from this sort of rate regulation around the same time the US did). That has nothing to do with cellular, which was always regulated under a totally different regime, and was built more than a decade after these events.
In the case of limited bandwidth, the "fairest" way would be to enforce a moving average cap. The most economically efficient way would be some sort of dynamic price system[1].
>But since people often have little or no choice over which carrier or ISP they use and switching can be expensive
In the context of this thread (wireless carriers), isn't this moot? I get how there are local ISP monopolies for residential internet, but the whole country has good access to all the major carriers.
[1] instead of getting 5GB, you get 5,000,000 "bandwidth credits". If a cell tower is below capacity, each megabyte of transfer only costs a nominal amount of credits (200). But as capacity approaches 100%, it costs more and more (up to 10000, or more). We could even have a total free market similar to bitcoin network fees where customers can enter a bid. If your bid isn't high enough you're limited to 2g speeds, or get no data at all.
Is there a reason why this can't be done? Have a carrier app that you can use to check the current rate, and that sends push notifications when the rate exceeds a rate that you've set. You can even do this through SMS so it doesn't take up any bandwidth.
>and allow me to accept or not?
Allow the user to set a maximum rate. If the current rate exceeds that rate, cut off data access. Allow the user to reactivate data or change the maximum through the app or through SMS.
>The flexible invisible cost is just not a good thing for end users
I never said it was the best approach, just the most efficient approach from a microeconomics point of view.
That's why I said switching can be expensive. If you discover bad internet speeds a month after you signed a standard 2-year contract with a carrier, what are you going to do?
Even if you pay that, you may need to pay for a new phone because of incompatible networks.
Not to mention the PITA of transferring your number.
(Similarly even if you have multiple ISP's, you may need to pay $$ for installation, a new modem, etc.)
Stuff just behaves too strangely for it to be anything else.
One video provider will get slow. Switch to the mobile connection and it's fine. Switch to another site on the wired connection, that's fine too.
They've made it perfectly clear they're unhappy I won't pay them an extra $100+/month for 900 channels of crap over and over.
Your ISP is purposely providing a too-small link to your video provider. So when a lot of people are using that provider, it gets slow. When you switch to another provider, or a VPN, you get to use a fast connection to that other VPN that isn't overloaded, and then a fast connection back to your video provider.
This way your ISP can claim they are not throttling traffic because they aren't inspecting the packets for traffic shaping, they are just using physically poor connectivity to certain providers for traffic shaping.
So these ISPs should be bearing the costs of peering to backup providers, correct, by your logic? Or would that be the backup providers holding ISPs to "ransom"?
Apropos of anything else, ISPs (generally) provide massively asymmetric pipes. You can't provide an asymmetric pipe and then complain that data use on that pipe is, well, asymmetric.
This idea is nonsensical. What if I want to go to somewhere that is three times removed from my ISP. Comcast has zero capability to force some South Korean ISP that is providing services to some website to increase their bandwidth with their upstream who is peered with whoever owns the undersea cable that is peered with the company that peers with Comcast.
The alternative is that Netflix asks for 40 Gbps to Comcast and suddenly Comcast is obligated to provide 40 Gbps to every one of its customers.
What we've always gotten with an ISP is a given link rate to our ISP. Beyond that it's always been a crap shoot.
The end result was Netflix doing a deal to connect directly with Comcast, cutting out the middleman (IIRC).
But who's fault is any of this? I say you reap what you sow. Comcast has only themselves to blame when consumers are skeptical of them.
In this case Comcast is getting money from it's customers and then Netflix has to pay extra for the traffic Comcast's customers requested and paid for.
This "rule" is too simple to game, to have ever been a functioning rule. Netflix could have just patched their app to constantly upload lots of random data. This would have killed ISPs who often expect mostly-down traffic profiles, but it would have evened out "the ratio" so it theoretically would have won the peering argument. Of course, Netflix never did this, which proves that this so-called rule never really governed anything.
I had to chose Comcast when I moved because the alternative could only handle HD streaming but not 4K. And then I'm paying half again as much to be able to watch a pretty movie while others are using the internet, or for several of us to watch shows on our own.
Comcast is getting paid for bandwidth, plenty.
[0] https://drpeering.net/white-papers/The-Folly-Of-Peering-Rati...
I checked several of the largest ASNs and several smaller NYC-area ISPs on peeringdb and they are all "No" to the ratio requirement. Google doesn't care. Amazon doesn't care. Cloudflare doesn't care. There's half the Internet.
Demanding money to peer with Netflix is just trying to charge customers twice because they can. It's not normal, conventional, or good for the Internet. Netflix might be able to pay, but will your startup be able to pay? Probably not. And that's not the future we want.
Instead companies like Comcast are seeking to make profit from this.
If I were of a suspicious mind, I'd wonder if Rogers is pulling some backend shenanigans to discourage cord-cutting.
CEO: You know how we solve this, offer them more channels at a premium price. Which will show them 1 or maybe 2 shows they want to see.
https://www.bloomberg.com/news/articles/2016-01-28/comcast-d...
However, assuming we bypass the garbage tiering, there is a legitimate reason to do this. Page-loads are simple and fast. Every one loading pages is not much of an issue, as they typically load at different times for a brief moment and then are done. This means for a rated connection speed, you don't need as much actual capacity.
Video streaming (or other downloads) are constant. To stream video at rated speeds, you must build to provide most users that speed at the same time. Problem two here is that in cases of high traffic, video streaming is greedy. If it's not going at full resolution, it will suck up all available bandwidth it is given. If more is given, it will raise quality (by default settings, at least).
A sensible solution to this is to stream video and audio over UDP and have the router just drop UDP when TCP needs bandwidth (to a point). The point is, requirements are different. People sending e-mails shouldn't have to wait a minute to send because you really want to stream your netflix every where you go and in 4k.
Which reminds me, I thought mobile connections were for simple stuff. Light browsing, email, what-have-you. Even with "unlimited" plans, you're throttled after twenty-odd gigabytes of data per month. Why can't people just download netflix in advance? Failing that, I'm not sure why people expect to use a wireless internet connection in the same way as a wired one.
one way for government to provide usable markets is to mandate good information without setting standards.
NFLX is over neutrality for home internet (not sure about mobile), they're in every server cage
maybe neutrality is the wrong policy
or maybe ISPs shouldn't be in the content biz
Wireless also does truly have true limits.
Or is the issue rural locations which were never wired for cable because they all just got satellite dishes?
EDIT: In comparison, the same dsl provider on streets that have had their lines upgraded offers 100Mbps plans for about $30 less per month than the cable provider. So some homes actually do enjoy a little healthy competition, but it hasn't been enough to cause any changes in the cable plan pricing yet :(
I guess the main risk is that doing this might wake up the provider who would then finish their rollout to get those customers back, and undercut you on price. So it would need to be quick and easy to grab your box and move it to some other neighbourhood in that scenario.
These guys are doing that around the Boston area. They aren't in my area yet but I have friends who have signed up and had good experiences.
It's very expensive.
...or how much of an issue it would be to plug into what's already there, but hopefully it's not terrible assuming the existing provider would agree to let you in.
They won't let you in.
I guess the main risk is that doing this might wake up the provider who would then finish their rollout to get those customers back, and undercut you on price.
And they also do that. National ISPs can afford to lose money in one city in the short term but you can't.
A few players like this tried to use streetlights for this sort of thing with WiFi. Hanging a WiFi point on a pole is outrageous unless the city owns the pole and plays ball.
* Spectrum (up to 400mbps)
* AT&T (5mbps dsl)
I technically do have some choice, but given that one of these is close to 100x faster than the other, and the pricing is fairly similar...
I think wireless isps and 5g are our best chance at disrupting this.
There was NO cable or dsl. ATT always said DSL was available in our area, and that they would schedule the install, then 3 days later, the local guy would call, and say "oh, hell no, lol" and cancel it
Another option that might not have been available at the time would be taking the Netflix device to a different connection (work, the library, whatever) and then pre-downloading a bunch of the cartoons for offline viewing.
I wonder why the cable option is so bad? Is it just because it's a single coax entering the building shared by all residents? Where I am (Kitchener ON), I can get 200mbps cable.
I have family living in a motor home, and they deal with something similar at RV resorts; they're stuck with the wifi available to them there, and little ability to use an alternative (short of cellular).
At this point, we should have symmetric fiber with at sub 50ms latency and 100mbps down and up everywhere like we work to have water, sewage, electricity, and gas. Right now, all of our money goes towards maintaining an antiquated system for providing commercial laden garbage media.
While I agree that symmetric fiber would be better, my speeds are always consistently close to what I'm paying for and I'm one of those types of people that:
For example, this is representative of what I typically see: https://pastebin.com/wSt8fdeeI count myself as one of the lucky ones and YMMV.
Or, you know, democratic policy changes a majority of us all agree are sensible.
This has reversed the actual cause and effect. Zero rural customers are served by cable. Some rural customers have satellite dishes.
There are numerous reasons why cablecos don't serve rural areas. The obvious reason is the expense that would be required to run a mile of copper for each customer. Less obvious reasons include the fact that if there is no municipal apparatus to target for corruption, the cableco business model sort of falls apart...
You don't even have to visit a rural area to see this. There are lots of satellite customers in cities and that fact hasn't chased off the cablecos.
In my city the copper service had a significant 3 week outage that when a CO switch failed. They would rather attrit the CWA people and pay a fine than maintain it. Verizon’s position is to wait for direct subsidy or 5G in urban areas.
https://i.imgur.com/zGSklKi.png - The local loop length is 2 km in this case.
but for wireless most providers have dense coverage maps. I use ting which operates on sprint or t-mobile I think so there's even more variety if you include resellers.
Even when you don't apparenly...
Wired on the other hand has no excuse.
Congestion is a real thing. Mobile subscribers are now using significantly more data due to video streaming, and wireless carriers aren't expanding their capacity fast enough. 5G won't help for most of CONUS. Millimeter wave 5G, which will help, will only be deployed in urban areas. Mid-band and low-band 5G will only increase spectrum efficiency around 20%.
For now, carriers are throttling videos and introducing various deprioritization options. All the major carriers only offer plans with forced traffic deprioritization. For some plans, all traffic is deprioritized. For others, deprioritization happens after a threshold.
There isn't enough bandwidth in the US to support 240 million subscribers streaming Netflix 24/7.
- I feel like there should be some oversight or required transparency about how and when they're applying throttling. Can they throttle customers in under served markets all the time to save money? - They're currently throttling sites not protocols or users. Why? Did some sites pay them off? If they need to throttle they should be applying that process equally among sites. "But AT&T didn't slow Amazon.com's Prime Video at all" - I would be curious to see where bandwidth problems play out in terms of "carrier could build more towers, but hasn't" and "we've reached the actual peak of what's possible with current technology.
For the most part, carriers sell unlimited, then throttle heavy users. While it's not honest, they're also not upselling those users on super unlimited plans. This makes me think carriers genuinely don't have the capacity for all this video. Without an upsell path, if they had the bandwidth, there'd by no incentive to throttle.
Not at all. TCP already does its own rate limiting without external throttling.
Individually, cars brake on the freeway to avoid hitting the car in front of them.
Collectively on a busy freeway, that sets off a chain reaction that results in the familiar rush hour crawl.
Minnesota did a study in 2001 on the effect of ramp meters, and disabling the meters (ie, removing throttling) resulted in statistically significant increase in overall freeway travel times.[0]
Statistically multiplexed shared networks like mobile wireless face similar issues. For a single TCP session the only metrics that can be divined by the endpoints are round-trip time and loss. As the shared network reaches capacity, larger numbers of TCP connections all back off around the same time, but large flows are more aggressive at ramping up than smaller flow (email/instant messaging, etc) and can result in an effective breakdown of network usability. A network control that has visibility to multiple flows and awareness of capacity of the system can influence overall performance much more effectively.
Ideally mobile providers would be trying to shoot for maintaining uniform latency per flow, similar to what queuing strategies like CoDel[1] achieve, but that's likely beyond the CPU and buffering capabilities of their existing hardware. Lacking the perfect solution, it's human nature to move on to the next approach: managing the biggest problem. Video is usually easy to identify, and so it wins the "able to be managed" prize.
[0]http://www.dot.state.mn.us/rampmeter/study.html [1]https://en.wikipedia.org/wiki/CoDel
https://www.netcraftsmen.com/wp-content/uploads/2014/12/2012...
> T-Mobile US Inc. throttled Amazon Prime Video in about 51% of the tests, but didn’t throttle Skype and barely touched Vimeo
> Choffnes says Verizon can’t restrict his ability to publish research and the companies that support him don’t influence his work.
Yet, missing from this article: any %'s from Verizon
Many warned then that they it's foolish to think that they can get ahead of competitors this way, but they didn't listen. They had this coming.
https://www.theatlantic.com/business/archive/2017/12/netflix...
A big one: I often do what should be simple things data-wise, like tapping a link to “view article”, and am punished. Is it my fault that the obnoxious target site decided on its own to download and auto-play a huge video with sound in a floating window, none of which was requested or required!?
I have the right to squash that crap to save my data plan and battery (and sanity). That has meant installing blockers, and it doesn’t look like I’ll ever be able to stop doing so.
Then complain to your ISP if that doesn't match your advertised speed.
FWIW, even if you don't complain, it will help Netflix gather data that they can use to complain on your behalf.
There is no way to complain, but if we had some open, trusted services similar to speedtest.net to publish the effective speed it would allow people to make informed choices when selecting an ISP.
This is throttling by content, not by user.
All video content gets throttled so that the html5 video player negotiates a lower video resolution. t-mobile literally touts this as a feature.
Net neutrality is the belief that isps should only be allowed to throttle by user, never by content, all data should be equal.
Easy way to implement one to test with here
https://github.com/StreisandEffect/streisand
at 9 cents a gigabyte, it's definitely going to be pricey.
So that same 2 hour movie on Hulu may be as little as 1.25 GB - Yep, 1/1200th of the original…. Or as much as 5–7GB for Netflix or Youtube.
That said, I use secure encryption to prevent the ISP from viewing my browsing habits. Throttling VPN use carte-blanche because you don't know what's in there would be overkill.
It seems to me that it's best to just throttle based on allowing burst usage and treating all bulk downloading including streaming video identically.
The article doesn't suggest that this problem exists outside the US. Perhaps it does.
A cursory scan of the paper found table 4 which suggests that the problem is principally a North American one nonetheless.
And by QoS I don't mean limiting video even if there's available bandwidth, I mean limiting it if other people are trying to use their connections reasonably.
In part, because if the throttle all bulk data transfers over some minimum size, all the time, most people are only going to notice on video streaming, since videos are the most frequent bulk transfers.
Why is that a reason to force them to use less than their fair share of available bandwidth?
> In part, because if the throttle all bulk data transfers over some minimum size, all the time, most people are only going to notice on video streaming, since videos are the most frequent bulk transfers.
How is that a reason for anything, and what is it actually supposed to be a reason for?
>How is that a reason for anything, and what is it actually supposed to be a reason for?
I was unclear. I'm suggesting that carriers are in fact, throttling all bulk data, but reporting is focused on video. Allow a sizable burst of data at the beginning of a connection and then start rate limiting, and most people won't notice except for on videos.
This was probably the simplest, least noticable, least costly intervention to reduce peak congestion and provide a consistent experience across the carrier network.
Is it fair? Maybe, if applied to all traffic
Is it clear and transparent? No, clearly not.
Does it allow for the network to be used to capacity? No, not unless the limits were modulated based on current use, which doesn't seem to be the case.
Is it good for customers? Unclear -- to the extent you can watch good enough videos using less of your data quota, that might be a good thing; to the extent that you spend more time downloading updates, that's probably not good, but would need analysis of battery impact traded off with better push latency if the radio is kept on for other reasons. Also, to the extent that this throttling provides better availability in congested area, that's a plus for users.
Seems unlikely to me, but who knows ...
> Also, to the extent that this throttling provides better availability in congested area, that's a plus for users.
If you are being throttled, that's obviously not increasing availability in any meaningful sense, so, no, it's not. What would be a plus for users would be adding capacity. Not performing the contractually agreed service is the exact opposite of "increasing availability".
Overall, the problem with all of this is that it borders on fraud. There would be no problem with selling low-bandwidth plans, if that is the only thing you can offer at economical prices. Or offering the user the choice to throttle certain services so as to conserve bandwidth. The problem is that carriers kinda-sorta sell a service and then don't do what they kinda-sorta promised. And a big part of the problem is that everyone frames this as "trying to help the customer by limiting abusers", or something similar, when really, the only abuser is the carrier selling a service they don't intend to perform, because they would have to spend money to do that.
The important part is not whether some traffic is throttled, the important part is who makes the decision as to what gets throttled. If I buy 1 Mb/s mobile IP access, and then the carrier throttles my connection to 1 Mb/s, that's my decision. If I buy "LTE full-speed IP access", and then the carrier throttles my videos to 1 Mb/s with no option to opt out, it is not.
Adding capacity is easy to say, but can be hard to do. Where I live, the city council basically won't permit new towers; whatever licensed spectrum on the towers we have is the capacity.
Trying to frame this as a contractual issue is a losing battle. Nowhere ever did someone contractually promise you a meaningful speed to an unknown destination at a consumer approachable price. Certainly, it's not transparent, and that's reason to be upset.
For instance, if you are 100 Mbits over capacity, you could either:
(a) Throttle 100 users Netflix streams, which drop from 1080p to 720p
(b) Throttle email, it backs up causing hour-long email delays.
One is obviously less harmful.
That is actually the first mistake in your argument in this context. The bandwidth limits in mobile networks are first and foremost an effect of decisions by the network operators, not an inherent property of the world.
> you sometimes want to limit flows in a way to produce the least harm. Throttling different streams has different effects on user happiness.
Well, maybe. But I would argue that that is both (a) prone to erroneous valuation, if only due to conflicts of interest, and (b) almost always only the case to any significant degree in very artificial scenarios, not in the real world.
> For instance, if you are 100 Mbits over capacity, you could either:
> (a) Throttle 100 users Netflix streams, which drop from 1080p to 720p
> (b) Throttle email, it backs up causing hour-long email delays.
> One is obviously less harmful.
Primarily, that's a nonsensical scenario. For one, you are never "over capacity". That's what capacity means: The maximum speed at which you can transfer data. You never transfer data faster than you can. You are only "over capacity" in the sense that the link is congested, but that isn't a data rate that you are "over the limit", it's just that the queue starts to grow.
But also, your scenario implies that throttling email more than its fair share somehow would cause "hour-long email delays". Now, I dunno, what would that look like? 100 users using netflix, ~ 10 Mb/s each, for a total of 1000 Mb/s, and one user sending emails at 1000 Mb/s, which would then lead to the email sender being throttled to 950 Mb/s under fair sharing, which ... still would not lead to anything remotely like "hour-long email delays", despite being a totally contrived scenario?
It looks more like you just made up one bad thing and one not so bad thing, and then noted that the worse thing is worse. Which is fine. But it is completely irrelevant to the discussion if you do not demonstrate how those two options are actually options in the same situation, and how they, respectively, compare to fair sharing.
There are two types of protocols: those that adjust the quantity of data sent depending on conditions, and those that don't. Streaming video does -- if throughput drops, it changes video encoding to send fewer bytes. Most other protocols don't -- they just get slow but they still need to send all the bytes eventually. There's a solid argument for throttling the first kind when congestion hits.
There's a huge literature on this. Start at https://en.wikipedia.org/wiki/Fairness_measure if you want to learn more.
Which is about as vague as it can get.
To make things more concrete: What would a gigabyte of mobile traffic cost if network operators were to optimize for low per-GB cost?
> There are two types of protocols: those that adjust the quantity of data sent depending on conditions, and those that don't. Streaming video does -- if throughput drops, it changes video encoding to send fewer bytes. Most other protocols don't -- they just get slow but they still need to send all the bytes eventually. There's a solid argument for throttling the first kind when congestion hits.
Except: There really is not, especially not between customers, which is what this discussion is about.
It is not some random accident that video streaming happens to be the thing that adapts to available bandwidth: That is precisely because it eats so much bandwidth compared to everything else. Which also means that throttling it in favor of other applications between different customers is actually not really useful. If you have one customer that watches a video stream and one (comparable, end-user) customer sending emails, then the email user will pretty much always have lower bandwidth requirements than their fair share anyway.
Also, it simply is not up to you to decide what a customer should value. If you sell a customer a contract that promises a certain bandwidth that is sufficient for high quality video streaming, then it is your job to provide that bandwidth at any time, except for rare occasions when/where unexpected demand hits you. It is not up to you to decide that a low-quality stream ought to be good enough, or that someone else's emails are more important than their high-quality video stream, if they are paying you for the same level of service. If the customer thought that that would be good enough, they would have bought a lower-bandwidth plan. If you don't offer a lower-bandwidth plan, then that is your fault.
Congestion should never be a standard condition in your network, if it is, what you are doing is essentially fraud by selling people a service that you don't intend to perform. The fact that it might be technically impossible to perform the service doesn't change that, it's still your fault if you know that you can't perform the service, but you enter into contracts anyway. Congestion should be an exceptional condition, and as such should never be so massive as to have any advantage from using application-dependent throttling between customers.
Why do you think that a user that is using exactly the same amount of bandwidth as any other users (i.e., their fair share) is abusing the system?
What would you say if some user was watching a video stream that used only half as much bandwidth as their fair share? Would that still be abusive?
You're arguing against the laws of physics here, and you're arguing that stupid users should be able to screw everybody else. No thanks.
You can watch video at 500 kbps perfectly fine.
> Dozens of users trying to watch video at 50 KB/s aren't even watching video; they're just launching a DDoS attack.
So, you are saying the problem is with people who can't watch video, but do watch video anyway, despite the impossibility? Or are you saying that people who can't watch video, and therefore don't watch video, are congesting the network by not watching video? I can't really figure out what your claim even is.
> I'd be very happy to throttle them to 1 KB/s so I don't have to wait 10 seconds to receive a picture through Whatsapp.
OK, I understand that you would like to get preferential treatment. That's fine, but does not particularly convince me.
> You're arguing against the laws of physics here,
Which law of physics exactly?
> and you're arguing that stupid users should be able to screw everybody else. No thanks.
How is getting the same amount of resources as everybody else for paying the same price as everybody else "screwing everybody else"?
Really, it seems to me like you are simply repeating a completely unjustified claim over and over in different wording ... you might want to consider that repeating a claim does not make it more convincing, especially if you don't address objections.
i think there is a misunderstanding here. he's not arguing that ISPs shouldn' be able to bandwidth-discriminate based on customer. He's arguing that they shouldn't be able to discriminate on the type of data. what would it matter to the other users if they were slowing down the network by html5 video, VNC, or playing a game over TCP?
I don't know a lot about networks, but i think a potential solution could be to discriminate between high-bandwidth and low-latency transmissions. For a video game or something you would want to have low latency, but you wouldn't need a lot of bandwidth. on the other hand, video streaming requires a lot of bandwidth but doesn't depend so much on latency as long as it has a big enough cache. images are probably somewhere in the middle. this would allow low-bandwidth applications to be prioritized.
That is a completely different kind of "prioritization" than what we are talking about here, though.
In particular, this "prioritization of low-bandwidth applications" does not in any way need to be aware of applications. All you need to do is to do fair link scheduling over all customers on a given link with some sort of token bucket per customer that allows short bursting. That way, if there is a customer that's been idle for a few minutes, say, they'll be able to transfer at, say, ten times the fair average rate for a few seconds, while other customers are slowed down proportionally, thus providing them with low latency. But that does not imply a higher average rate, nor a distinction of video vs. non-video, or anything like that. On average, every customer on a congested link would still be getting the exact same bandwidth, just as a constant low-bandwidth stream for some and as a series of short, fast bursts for others.
What tcxgy is arguing for is that if you are streaming video (or whatever other application that he doesn't like), then you should only be getting 2% of the average bandwidth compared to someone who is using an application he approves of. So, if you are constantly posting high-resolution pictures (something that he seems to approve of) at an average of 50 kB/s, that should get 50 kB/s, but if you are watching a 25 kB/s video stream of some sort, that should be throttled to 1 kB/s, because otherwise it's supposedly violating some laws of physics, and it's abusive to use half the bandwidth for video, and whatever other nonsense arguments he has come up with so far.
Default would be the first so that networks may only have special treatment for apps that explicitly opt in to say "i'm ok to wait a bit" like Netflix, Youtube and friends.
And you seem to be confusing physics for economics. It's the latter that determines available bandwidth in most cases.
So, you want to allow companies to not fulfill contracts because the company has failed to fulfill contracts?
Mind you, if customers in this scenario of yours don't get to eat as much as they can eat, that has absolutely nothing to do with the other customer that happened to eat as much as they can eat, and everything with the company not being able to fulfill the contract that they entered into. It is the company's fault, and only the company's fault, if they enter into contracts with multiple people that say "N bucks for as much as you can eat", if they then don't have enough food to feed all those people.