Slashdot Mirror
Stanford Engineers Propose A Technology To Break The Net Neutrality Deadlock (phys.org)
An anonymous reader quotes a report from Phys.Org: Stanford engineers have invented a technology that would allow an internet user to tell network providers and online publishers when and if they want content or services to be given preferential delivery, an advance that could transform the network neutrality debate. Net neutrality, as it's often called, is the proposition that internet providers should allow equal access to all content rather than give certain applications favored status or block others. But the Stanford engineers -- Professor Nick McKeown, Associate Professor Sachin Katti and electrical engineering PhD Yiannis Yiakoumis -- say their new technology, called Network Cookies, makes it possible to have preferential delivery and an open internet. Network Cookies allow users to choose which home or mobile traffic should get favored delivery, while putting network operators and content providers on a level playing field in catering to such user-signaled preferences. "So far, net neutrality has been promoted as the best possible defense for users," Katti said. "But treating all traffic the same isn't necessarily the best way to protect users. It often restricts their options and this is why so-called exceptions from neutrality often come up. We think the best way to ensure that ISPs and content providers don't make decisions that conflict with the interests of users is to let users decide how to configure their own traffic." McKeown said Network Cookies implement user-directed preferences in ways that are consistent with the principles of net neutrality. "First, they're simple to use and powerful," McKeown said. "They enable you to fast-lane or zero-rate traffic from any application or website you want, not just the few, very popular applications. This is particularly important for smaller content providers -- and their users -- who can't afford to establish relationships with ISPs. Second, they're practical to deploy. They don't overwhelm the user or bog down user devices and network operators and they function with a variety of protocols. Finally, they can be a very practical tool for regulators, as they can help them design simple and clear policies and then audit how well different parties adhere to them." The researchers presented a technical paper on their approach at a conference in Brazil.
How about just 'technology' or 'new technology' ?
Could we please get everyone to implement RFC3514?
https://www.ietf.org/rfc/rfc3514.txt
Please? It would make network security a lot easier to deal with.
This is like the "do not track" button.
Only worse.
Every advertiser in the universe will want to programmatically toggle this option "for the convenience of the user."
No. Treat all traffic identically. Bits from CNN are more more important than bits from lemonparty.com
Nobody gets special treatment, that's what net neutrality IS.
Idiots.
It's not about the users. The whole reason ISP's want to give preferential treatment to traffic is specifically so that they can force content providers to pay them for access to their customers. They want to pick the winners, punish competitors, and make money doing it. Anyone that thinks this is about improving the end user experience isn't paying attention.
In what way does this break the deadlock? It does not allow ISPs to charge content providers extra for access to their captive user base, so no extra revenue there. Worse, it allows users to deprioritize the spam content they don't really want where the ISPs are currently getting their extra profits. It seems like this is just good for users at the expense of ISPs, so the ISPs will never go for it.
Which will obviously give me an advantage over everyone else because they sure won't do so.
http://lkml.org/lkml/2005/8/20/95
Way back when, the definition of net neutrality was not "the proposition that internet providers should allow equal access to all content rather than give certain applications favored status or block others."
When I first heard the term in the 1990s, net neutrality meant that the main trunks all processed data the same for every provider and end user. They could certainly make the decision to route some data packets before others, such as video before text. The problem is that the ISPs are now also providers, and have decided that their video is more important than another provider's video. So Comcast is fucking with Netflix, claiming Netflix pushes out too much data. But if I am Comcast's customer, I don't want them disrupting my video feed just because they want more money than they already gouge from their customers.
If you think I voted for Trump because of this post, you're wrong. I voted for Dr. Jill Stein of the Green Party. Again.
Customers will be given a choice whether they want all of the internet at modem speed or whether they want a select few faster. What clever compromise!
Well, you know, the new rules and regulations we added ended up having all kinds of unintended consequences (that people warned about repeatedly, my goodness, who would have thought), so let's add yet another system on top of the existing pile of crap. Soon it will be just like a Microsoft product! Can't wait! Nothing says "Freedom" like more interference!
Love sees no species.
To 99.999999999999% of users, they really, really do in fact want traffic from CNN to have priority. Especially over your LemonParty traffic, never mind your own.
Trying to fight want humans fundamentally want is not only futile, in the end it is stupid and pointless.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
These engineers really don't understand humans. Every human, including corporations, all think their data is more important than everyone else's. Hence making everyone's traffic have the same importance is the only equitable way to run things.
ISPs wouldn't go with this simply because it'd require quite a lot of extra work on their end to make this happen for zero gain to them, and everyone knows ISPs do traffic-shaping for their own benefit, not for their customers' benefit. Also, end-users would just tag *everything* as to be prioritized, because they obviously don't want any of their traffic to be slowed down, so what would be the point? Besides, how the fuck would you even implement this for something that doesn't use a web-browser? Ask the ISP to list every possible network-protocol ever invented and all the ones still waiting to be invented, so you can click on them? That'd be one ginormous list to go through.
Also, I have to take offense at the whole "But treating all traffic the same isn't necessarily the best way to protect users." -- works fucking well over here in Finland, but then again, our ISPs aren't nearly as obsessed with overselling capacity. Maybe fix ISPs overselling their capacity, instead of trying to come up with workarounds that only harm end-users!
...whatever data request i'm sending right now.
That'll do, ISP, that'll do.
Warning: Teh poster of this messaeg is lysdexic
A user sets they open source game, email, p2p to use the best Network Cookie speed by default every time?
Hardware is sold with a best Network Cookie speed always on setting? That would set every user on that network at top speed just for buying a new router.
i.e. the user-directed preferences was to buy a new router that sets the Network Cookie to max for every packet.
So will the providers then be allowed do deep packet inspection and be allowed to guess that email, an open source game, p2p will not be getting that "fake" Network Cookie setting after all and slow things down a lot?
Other packets will get set to max speed all them time, if they pay the networks.
Better just to let the user select from a really good provider or low cost over subscribed provider.
Then every packet is only limited by bandwidth, not later slowed by some cost cutting "Network Cookie" setting that some network or telco later flags many packets with because they can.
The end user will be left clicking "Network Cookie" settings all day only to have some distant telco apply their own rules to packets that have no expectation of speed on their network. Preferential delivery will need an extra payment.
Domestic spying is now "Benign Information Gathering"
Wait, didn't something similar exist in the past, call Type of Service field?
https://en.wikipedia.org/wiki/...
Of course its been deprecated, for obvious reasons.
I want all providers to be locked to the same standard.
If I was dumb enough to have the top package that comcast offers at the top speed - they should be limited by the cap they have on their lowest speed.
They should also be forced to charge the same "internet only" price to their customers that have cable television in addition to the cable TV only charges.
Finally, they should be forced to offer cable tv only at those same prices.
At that point cord cutters won't be forced to subsidize cable tv.
_ _ _ Go for the eyes Boo! GO FOR THE EYES!
From my vantage point, the public wants net neutrality and the cable companies don't. If the cable companies want to give preferential treatment to content providers that pay them off, they should lose common carrier status with all the liability that entails.
Set out with an aim to provide a 'make everyone happy' solution for the net neutrality problem, according to their paper.
They failed. This is open to massive abuse and i'm not happy about it in the slightest.
This basically sounds like QoS, where if there is network congestion, certain traffic can be prioritized over other traffic. If there isn't congestion, I honestly don't see what the point is to this besides to get funding to develop this "ground breaking" technology from investors. The entire reason why ISPs want to break net neutrality is to get additional revenue streams from content providers to make their services more enticing over competition to the eyeballs served by the ISP. The description of this technology seems to violate the entire reason for net neutrality to be violated in the first place.
Network Cookies allow users to choose which home or mobile traffic should get favored delivery, while putting network operators and content providers on a level playing field in catering to such user-signaled preferences.
This is not what network operators and content providers want. They want control over what users can do and see. Also, they don't want a level playing field. That's why we're seeing all this dancing on the line of 'net neutrality', folks are testing what they can get away with. It's a novel idea, but I personally doubt it'll ever leave the 'drawing board.'
Wait wait, so they are effectively asking end users to QoS themselves, and transit is supposed to honor that? What happens when users QoS flag all their shit to realtime? The claim of signatures preventing transit or middlebox tampering is hard to swallow as well.
This could only work if "preferred" packets cost more than other ones. Otherwise users would set them all as preferred. With this only wealthy people would do that....
there is only one solution. internet service providers must be BIG DUMB PIPES. there is no debate, there is no compromise. transfer the fucking bits without bias, without throttles, and without caps (and without sniffing, snooping or injecting shit into them).. period. at least for wireline services now. wireless, too, when next-gen (aka '5g') system is deployed with substantially higher capacity.
if, you, a provider faces congestion issues, it's your own fucking fault for overselling your resources too much. control resources not by caps or throttles but by the speed of customer access plans, and expect them to, i dunno, actually use what they're paying for, because they will do exactly that. and allow large services, like cdn or netflix to hang cache servers off your network for *free* (they provide the hardware, of course), because, ya know, it saves you money and keeps your customers happy; and they'd be happy to do it because it saves them money and keeps their customers happy too.
I just read the paper. They have some ideas about implementing QOS (quality of service), but as others have said, nothing they've presented actually has anything to do with the real issues around net neutrality.
One well-known offering related to network neutrality went like this:
If the user allows videos to play at a lower resolution (which is much lower bandwidth), the service provider would make that bandwidth free, it would be exempt from mobile data caps.
A mechanism for the user to choose which traffic is free doesn't do the trick because the offer is an EXCHANGE. The ISP is offering the video bandwidth for free IN EXCHANGE for the user accepting lower resolution.
Newflix, a new company wanting to compete with Netflix, might say "rather than our customers paying for a higher priced plan in order watch our service all night, we would like to pay the extra cost and subscribers with even the el-cheapo internet plan can use our service, because we'll pay the extra cost direct to Verizon." It doesn't work to have the subscriber say "I want xvideos.com to to be zero-rated", because xvideos.com" isn't offering to pay the difference, Newflix is.
And yes of course these kinds of deals can also turn out to be bad for the consumer; they can also be good. Personally I'm on a cheap mobile plan, it costs very little and after the first X GBs each month, it's throttled slower. My phone is small, so 480p Youtube is fine for me. I would *like* to have the option of zero-rating youtube videos in exchange for watching them at 480p.
Bullshit.
Net neutrality is supposed to protect the users from the networks throttling data users want in favor of data the networks want. (i.e. ads having higher priority than real-time gaming or web traffic, because the network gets paid more for ad traffic for example). But a user preference system does not solve that problem because the user can only specify preferences for his own little segment of the network, namely, the uplink between him and the provider. The network providers can still throttle traffic all they want and the user's preferences don't mean diddly in the big picture. Even if the user's preferences are set and somehow considered in routing on the wider internet, basically everybody would mark all their packets as high priority, so it's the same as having no priorities. Unless high priority packets cost more. Which is exactly what net neutrality is trying to block from happening. Suppose you use Bit Torrent. If you set preferences to prioritize Bit Torrent traffic on your uplink, but the networks throttle Bit Torrent traffic on the trunk lines and backbone, guess who wins? All other traffic. Your preference settings are irrelevant in the big picture because the performance is dominated by the load and traffic shaping on the other networks your packets travel though. Although sure, you could prioritize certain traffic over other traffic on your internal network and outgoing on your uplink. Oh wait, you already can, in almost all consumer grade network routers....if you look in the options they allow you to set QoS and prioritize certain types of traffic. But none of them can control what happens on the uplink side or out in the network past your provider.
Slaves had the choice of being fed regularly or fending for themselves.
Time is what keeps everything from happening all at once.
The whole net-neutrality issue is moot; near as I can tell, there is more than enough bandwidth bandwidth to go around, when normal packet prioritization is utilized.
The real issue is the con-artist ISPs trying to double-sell the same service, charging a premium to both sides.
Mod me down with all of your hatred and your journey towards the dark side will be complete!
Every corporate network or ISP will de-prioritize external traffic and every web-server will prioritize its own traffic. It will be a war over the speed setting.
This could be really cool if I could flag ads as a low or zero priority while retaining the other content as high or desired priority. On the other hand what if I don't want to flag any content as lower priority but 'want it all' delivered at the highest rate or the advertised rate. It seems that this could be interpreted as what stuff can we slow down, not what do we prioritize.
errr....umm...*whooosh* *whoosh* Is this thing on ?
I mean, you are literally saying that some traffic will be given precedence by request. That's literally the opposite of treating all traffic equally.
No! It's a *SIG*. Keep the Special Interest Groups away! (Con joke!)
TOS was replaced by DSCP [https://en.wikipedia.org/wiki/Differentiated_services] which offers finer granularity of control.
There are a number of classification, prioritisation, shaping and policing technologies applied at different layers of the network within a Quality-of-service (QoS) framework: This is almost a black art in the Network Engineering field it's so complex & such a rapidly changing area.
Trouble is you can't trust someone else's DSCP so QoS is implemented only within network boundaries and with trusted peers.
The base problem with anybody being able to specify their traffic importance is that everyone thinks their traffic is important: Torrents are an embodiment of this: start as many connections as you can to get the data as fast as possible: Don't mind that you're sucking more data than your last mile link can possibly deliver.
Isn't this just QoS? I guess if its really user friendly and done through a browser (or an easy to use network configuration panel on your computer), and if it has some bearing on upstream nodes, not just your own router.... but still.
If we assume that ISP is neutral, why this can't already be done by client side QoS? Limit you torrent and your web will load as if nothing is hogging your bandwidth, just don't saturate your link and prioritize traffic you care the most! Why do you need to make existing stuff so complex?
Cisco has had the necessary technologies for some time - RSVP (Resource Reservation Protocol) and PfR (Performance Routing).
RSVP is used for multicast streams today. It should be adapted for the use of unicast. Note: Streams are typically the only type of traffic where priority is critical (voice streams/video streams). RSVP basically negotiates available bandwidth for a stream on the given path.
PfR is used for intelligent routing of traffic based upon the needs of that traffic. Think of routing tables based on traffic types (voice, video, web traffic) instead of subnets or autonomous systems. Traffic is routed where it will be most successful. Interesting stuff. As one might expect, this involves more device memory.
In my opinion, really what is needed is both of these technologies combined. PfR-RSVP, where PfR decides best route, and RSVP reserves necessary bandwidth, or works with network devices on the way to negotiate what bandwidth it can currently offer. I could say so much more here, but sometimes ideas left open work better.
The two protocols above are used in Enterprise wide area networks today. In fact, Cisco has coined the use of PfR+DMVPN (internet based VPN MPLS) as "IWAN" (Intelligent Wide Area Network).
Please note that I own none of the trademarks above, nor do I claim to.
Read the paper: The ISP provides a list (on a well known server) that the user chooses from. That means ISP will offer 6 sites. We get that now from ISP's (for example T-Mobile). Nothing
Oh GODS, here we go again! Neither this nor any bundle of rules and laws can legitimately be called network neutrality. How has the discourse about about this crucial topic been so completely co-opted and misdirected?
Network neutrality is what happens when citizens collectively own the network infrastructure, not the various builders of bits and pieces of it. It's shared infrastructure, just like roads and highways: do we allow the builders and maintainers of those to retain ownership? No, they are contractors for a public trust.
The Internet is not different, yet certain vested interests have managed to divert attention from this and misdirect the conversation to techniques and tactics which they are already quite skilled at thwarting. The service providers are lying to you: as much as they claim to despise and fear network neutrality LAWS, those are exactly what they do want. What they do NOT want is any conversation that suggests a public buyout of the wires. As long as they physically control the wires, network neutrality cannot and will not ever exist.
themselves above my data? This is utter crap like all the self-professed experts who think they know how to do scheduling. This is a scheme to torpedo net neutrality by the baffle-them-with-bullsh*t method. This kind of crap should be on the list of topics disallowed by papers until the submitter makes his first billion, and heavily peer-reviewed then.
And it already exists, and for most operating systems, requires no added code or features in order to implement. The only issue is getting applications that don't already utilize it to include it in their packets, and for some people getting all hardware out to their wan port including QoS routing functionality to ensure packet delivery according to the specified priority.
> You can already set your torrent client to self throttle. The ISP does not need to do it for you, and should not be in the position or business of doing it for you.
You can reduce the *bandwidth* of the torrent or whatever between you and the ISP. You can't do crap about how it's treated for the other 99% of the route, currently. And reducing the torrent bandwidth isn't actually what you want. You're hoping that reducing the torrent bandwidth will have the side-effect of reducing the jitter or at least the latency of your voip traffic. That is by no means guaranted.
There are three different measures of connection quality, each important to different applications. For torrent, you want max bandwidth, you want to transfer as much as possible every ten minutes. For voip, you only need 64Kbps, but the main thing is that the latency be consistent. That's called jitter. You don't want 5ms latency on one packet and 25ms on another, because they would arrive out of order, and it would sound like you said "uto of roedr".
For Netflix, you want a consistent X Mbps, with the same amount of MBs transferred every ten seconds. You don't mind if it's fast for one second, then slow for one second - Netflix buffers.
The engineers at the ISP know about this stuff, and can control their network to give best results - low bandwidth low jitter for voip, high bandwidth high latency high jitter for downloads, medium bandwidth high latency medium jitter for streaming video, low latency high jitter for gaming, etc.
> On the other hand what if I don't want to flag any content as lower priority but 'want it all' delivered at the highest rate or the advertised rate.
There are three different measures of connection quality, each important to different applications. For torrent, you want max bandwidth, you want to transfer as much as possible every ten minutes. For voip, you only need 64Kbps, but the main thing is that the latency be consistent. That's called jitter. You don't want 5ms latency on one packet and 25ms on another, because they would arrive out of order, and it would sound like you said "uto of roedr".
For Netflix, you want a consistent X Mbps, with the same amount of MBs transferred every ten seconds. You don't mind if it's fast for one second, then slow for one second - Netflix buffers. (Cable modems naturally do exactly this, fast-slow-fast-slow-fast-slow. Good for Netflix, bad for gaming and voip.)
The engineers at the ISP know about this stuff, and can control their network to give best results - low bandwidth low jitter for voip, high bandwidth high latency high jitter for downloads, medium bandwidth high latency medium jitter for streaming video, low latency high jitter for gaming, etc.
Who paid you for this? You know better. Lier's you know this is all about setting up the user for tier payments. The rich get the goodies and the poor get the shaft! .
Originally, Net Neutrality was the response given when some of the networks came out with a plan to start double-dipping and charging sites for access to users, rather than just charging users for internet. Everyone supported it. Everyone. Right, left, everybody saw that as a transparent money grab and told the people trying it to GFY.
But... then politics slipped in. Some on the left thought it'd be a great time to get more internet regulation, and some of them even honestly want that to help. Some on the right wanted more freedom for corps to screw us while others were nervous about regulation. So the lobbyists are doing their best to divide our interests along party lines, because back when we were all united, they couldn't get away with this crap.
I just wish I knew how to unwind things and get us all to remember being on the same page. Nobody likes this double-dipping crap, except those exploiting us all. The only real question is whether we can agree how to stop them...
Here's a way to break the deadlock. Those that want to discriminate traffic and charge extra for fast lanes, fuck off and start a new internet using all your own infrastructure. Or just fuck off and don't bother with the rest. Deadlock broken.
Wanna buy a shirt?
https://www.redbubble.com/people/stealthfinger/shop?asc=u
In the physical world this is done by giving visitors the possibility to pay to jump queues.
It is nothing more than an attempt to monetize congestion, therefore removing any incentive to eliminate the congestion.
The dark fiber will stay dark.
There are a few intellectual groups who think the answer to every perceived problem
is a political, governmental solution- this will only result in more inefficiency, as the market is restricted
by governmental regulation...raising prices for the consumer.
Something I don't get is why would anyone want another tool to "configure their own traffic", this time ISP-side, when clearly there are already equivalent QoS control tools, which don't even require leaving the network boundary of your residence for their use. Implementations of software or hardware-based QoS might not all be straightforward, but that goes to show just how useful most people find "individual traffic rating". It's very niche. Let's face it: most people that pay for a connection want it full throttle, 24-7, no restrictions, because most people change their browsing habits fast enough they rarely need QoS control. What they also expect is what they pay for: a dedicated line for whatever needs they have - not a customized data plan that will have (e.g.) faster video streaming and slower torrent downloading.
What this tool would bring, in practice, is a platform much like the default-on, opt-out-type state/isp-sponsored parental control implemented ISP-side in the UK, and I believe 200% the only real reason research is being done in the matter is that some form of legislation can go into effect which will force users to get some "standard QoS" turned on by default, then be able to opt out obscurely enough to prevent most - which is what ISPs want, to block the ignorant majority. This is gonna sound super-commie, but the best way capitalists have of making money is gently penetrating every ignorant sucker that fails to look at the rear-view mirror. After something like this is set in place, it becomes a matter of the ISP's and the gvm't "loopholing it" to having something much like the great Chinese firewall, and people getting "neutral" shit blocked because "state says I should not like it".
Given that, let's focus on the real, and only problem of net neutrality - it is on a grand scale, NOT individual. A simple example is when other uplinks, or clients of an ISP, are hampered by the 1GB plan guy down the street who hogs the neighborhood by running a private usenet server, or by going on a 4k Netflix rampage. An entire city failing to submit their taxes because everybody is downloading/streaming a new episode of Game of Thrones. Stuff like that. You can't control this with something like what the Stanford guys are doing: the 1GB plan guy pays for what he has. The problem is ISP's offering something they can't actually provide, becoming FSPs - failure service providers.
It seems that these researchers have a fundamental misunderstanding of the underlying causes of this debate. ISP's want to sit in the middle between businesses and users and charge both sides as much as possible to talk to each other. They don't care what we want prioritized, they only care who will pay them to be prioritized.
They might as well have made an app that tells Donald Trump when people want him to shutup. He doesn't care, neither do the ISP's
Whats is wrong with "A" technology? It is a technology if it solves a problem.
This appears to be a technological solution to a social problem and those rarely work well. Net neutrality is only a problem because certain companies feel their economic self interest should be more important than the good of the overall system or the needs/wants of the end user.
I actually agree with what this solution does. No one will care that their netflix packets are prioritized lower than their voice packets, since netflix streams and voice needs to be near real-time. Same thing for SSH sessions, page loads, or IM applications. They need faster response times than your Carbonite subscription or drop-box sync.
They will care when AT&T or Comcast starts a massive campaign to convince people to prioritize the services they favor over the ones that the user might otherwise choose. What, you think they'll sit idly on the sidelines over something that could make them huge amounts of money?
Doing it this way (but making it adjustable to the home user by doing something like... right-click on the application and set its "priority" on a scale or something) could be really useful, especially in bandwidth-limited deployments when your backup starts and kills your phone conversation.
This will fail the "mom test" horribly. I can see the family tech support calls coming in now. Shudder...
Not true. If I am audio chatting with a friend, I want packets delivered in milliseconds. But if I am running a torrent in the background, anytime in the next hour or so is good enough. It would be nice to be able to set my own preferences.
No it really wouldn't. Maybe you are technologically adept enough to make sane decisions. Most people are not. My mother doesn't even know what a torrent is much less what priority it should receive. Furthermore the moment you put a control like that in, companies with an economic interest in seeing their traffic prioritized will bend heaven and earth to get it adjusted in their favor. Don't for a moment believe that they would not.
Basically you are naively thinking this would be some innocuous control used only by well informed end users with economically interested companies politely deferring to whatever the users want. The real world doesn't work that way.
Any implementation of this would require the global consent of the ISP "community", it used to have some "community-ish" behavior but now the bean counters are deep into the routing.
Not going to happen:
1. It sound like a major attack vector, to ddos backbones etc. We in the ISP community are real hot about not taking risks with the network.
2. It sounds like a major demand on how QoS is implemented in and between networks. This is something ISP do not agree on today and do not trust there peers with.
3. If it requires software updates it will take 1-2 year to be accepted as stable in the "community" and after that it would require 2-5 years to reach maturity with end-to-end functionality.
4. Settling the discussion. It would take years to discuss what the user would be allowed to do and how that should be handled in different scenarios. And what the ISP must accept, and run as default. Like preference to high income sources.
All the above will take so long that we will all have Gigabit internet and massive backbones in place before this will allow ISP to run there backbone overloaded with the excuse that customers can choose traffic priorities.
From what I gleen from the abstract this would also be a massive load on all the transit routers. Far easier to keep doing what we have been doing, just keep upgrading.
Net Neutrality is terrible policy. For the one issue it purports to solve, it just moves the ridiculous pissing contest to the peering point. It then ducks up everyone's network but good. You are stuck zero rating management traffic on the sly so that customers don't complain about the mismatch between the bits going into a gateway and what is reported on their bill. Think your ISP has no good reason to prioritize DNS traffic? Think again! Duck all of you who have never tried running a network with users like you on it for ramming this shut down our throats. Don't like your ISP double dipping? Let them know or make that the real FCC target. Don't like your ISP using policy? Take a deep breath and ask yourself why you care. But ducking network bits do not have ducking God given rights to non discrimination.
This post auto-censored by my non-exploding Samsung phone.
The entire business model converts disk space into repeated bandwidth usage. The whole reason we have these discussions is because media companies haven't figured out a way to stop people stealing the stuff they stole from other people.
My ism, it's full of beliefs.
The real issue is not being talked about.
The issue is service providers wanting to separately monetize something that is currently part of a package. This is just a strategy to increase profits from something they already provide.
I am not paying for partial access to the internet.
I thought is was bad enough when service providers quit hosting newsgroup servers.
Service providers expected profits to go through the roof with the expected increased volume of subscribers, the short sighted bean counters predictions of ginormous profits fell short as usual when demand for bandwidth outpaced the volume of new subscribers, rendering that 20 year old equipment obsolete.
So instead of going after the cause of the high band width usage they are going after the end user because it is easier and more profitable to extract a little from a lot of subscribers than it is to bill the Facebooks and Youtubes of the internet. Then while they are leaching all they can from the subscribers they will double dip by putting the squeeze on big traffic sites to pay up. More $ for less product, its the Corporate way "New and Improved".
Rick B.
The paper presents a technical solution to a problem, but doesn't state what the problem is. It pays lip service to network neutrality, but demonstrates no understanding of the actual problem. If you allow users to choose what sites to prioritize, a logical user will choose "whatever site I am visiting now." If you ask them which sites should not count toward their data caps, they will answer "whatever site I am visiting now."
This is like having a special ticket that you hand to a cashier that tells them which items in this shopping trip you want to be free. They will obviously pick the most expensive item. They can also choose which lane they want to run the fastest. They will obviously pick whichever lane they are on. The solution is entirely unworkable.
Giving users the ability to choose this doesn't do anything for the ISPs. It also doesn't do anything for the users because it just means they picked which sites to slow down. Nobody wins here. The only incentive to do this would be to confuser users into thinking they have some kind of choice for marketing purposes. There is no material benefit.
Would someone please explain what zero rating is to those Stanford engineers. They obviously don't know it. It's an accounting product, has nothing to do with QoS or fast-laning IP traffic.
Then subscribers make everything 'top priority' and nothing is improved; those who actually implement low-priority channels, suffer.
Make the routers smarter: A computer instructs the router which socket/port is the priority channel; the router implements a timed lock-out so the computer can't issue a 'change priority' instruction before each data packet. The ISP/backbone can honour the priority channel scheme depending on the plan purchased by the subscriber.
Are these Stanford professors similar to the Harvard professors who were paid by the sugar industry to show that sugar was good for you but fat was evil?
There should be the "internet" and then there should be private networks on the side for prioritization. They should physically be different networks. Implemented kind of like how local and long distance were 10 years ago.
Problem there is getting everyone to cooperate on the "private" network since it will be free game and everyone will want their cut.
I am on skype with (significant other/GF/BF...) and they are asking hard questions, can I get a button to slow down my connection speed to a crawl ?
How will this technology enable ISP's to blackmail content providers and double-charge customers?
Slashdot social media options: AIM, ICQ, Yahoo, Jabber and Mobile Text. Why no MySpace?
Prioritize all my traffic. The only people who will have a problem with this are those who don't know how to prioritize their traffic. This creates a "The haves VS. the have-nots" situation, which is beneficial to society by creating frustrations, schisms, anger, delays, and potentially loss-of-life as e911 systems aren't prioritized by those who don't know how. Or, you know, we could keep the Internet EXACTLY as-is, which fosters incredible growth, economy, and anyone has the chance to be the next big thing
Despite the greed of predatory capitalism at its ugliest shown by ISPs, there is a real bandwidth problem with mobile.
Which has nothing to do with net neutrality. That is not a valid argument for Comcast to be allowed to prioritize their NBC data over data from Google.
A fair, metered, user-controlled, pay for what you use, is the answer
Already have that on wireless. Basically everyone pays for some amount of data per month. How this data is prioritized should not be up to the wireless carrier. If I use more then I pay more. On wired it is a non-issue. There is no spectrum limitation there.
but in practice, providers will price the various levels and speeds of data transport the same way they price cellular and cable plans. That is to say, the pricing models will be utterly arcane, difficult to understand, obfuscated to the nth degree, and designed so as to make comparisons almost impossible. And then there will also inevitably be the same kinds of 'inconsistencies', (to give the providers the probably-undeserved benefit of the doubt), between the usage recorded by the user, and that recorded by the provider.
Network Cookies sounds like a technically useful development; but the inventors' naive view of how it would be a fair substitute for true net neutrality has 'egghead fail' written all over it.
'The Economy' is a giant Ponzi scheme whose most pitiable suckers are the youngest among us and the yet-unborn.
That famous offer you were referring to could be taken advantage of by any company that used the auto-throttle video protocols that the provider supported. As it happens, these are pretty common protocols most of the big boys (YouTube, Netflix, Hulu, Amazon Video) were already using, but you could have filled out the form promising the various guarantees of the video and gotten "raymoms's awesome videos" zero-rated too.
That's why it didn't violate net neutrality. While I would have loved to have an auto-detection of that protocol, the fact that it was free to the providers made it (in my opinion) net neutral.
And there is literally nothing in a net neutral world that stops Newflix from doing what you're suggesting. What the net neutral world is stopping is Newflix paying for that different tier to appear only if they are connecting to Newflix.
Your ad here. Ask me how!
Seems like somebody is struggling to define NN to be something it shouldn't be.
This article is not about how to do net neutrality.
It is about a tool for implementing network management that could be use for or against NN.
The tool is as follows:
A user asks a network provider's server for some 'cookies'.
The user adds these to some of his packets.
The network sees this tribute and places the packets in the fast lane.
The theory is that this is NN because if the 'user' is choosing which packet get special treatment, then this is neutral with respect to content providers.
In practice, it seems to me that the user is already choosing which packets are important to him when he chooses where to click.
This just adds another way to collect cash for the internet service he already paid for.
Also, how long before special deals are cut so some content providers have an good supply of cookies. (zero rating?)
To work technically, the cookie server needs to do connect admission control (CAC) to make sure that there is congestion free bandwidth available to make the fastlane mean something useful. This means that cookies need to be for specific source/destination pairs for specific times and rates. The server would have to take this request plus the current network topology and decide what to do. I fail to see how implementing such a server is possible at scale if this works out for more than a select few.
The problem with the Internet today is that it is a congestion free for all. A user's access to a congestion point is determined by how many times he tries. This is bad because human nature makes for excess trying which makes the overall situation worse. If this issue could be addressed in a fair manner, the NN would be a non-issue. This seems the best path to NN.
This tool might get us there if all packet need cookies and the cookie allocation policies are fair and work at scale.
Such a scheme seems a hard way to accomplish this because it puts too much state in the concentrated places in the network.
Putting the state in each packet seems a better way to accomplish fairness.
Rainbow queueing has been around for a long time, but it requires modifying the per-hop behavior of each router.
Until something like this is done, I don't see the struggle sorting itself out.
While this is a great technology, this doesn't solve the actual problem.
The whole reason Net Neutrality is even an issue, is because of corporate greed. ISPs (at least the big ones) want to be able to double-dip by charging both their customers AND content providers for using their network. They can't deny traffic outright to entities that don't pay, cause that would be universally considered to be a Bad Move(tm), but they feel that they can get away with the whole, "That's some good data you have there. It'd be a pity to see it slow down."
That's the one and ONLY reason Net Neutrality is even an issue. The internet *already* had what amounts to Net Neutrality, baked into it's system from day one.
Anyone care to explain why people wanting to stream the 4K version of Plan 9 from Outer Space should be given the same priority as real-time medical imaging data?
there are a few examples of where preferential routing should be considered, but mostly the telcos will use it as leverage to earn more money for high-speed lanes.
Remember kids, if you're not paying for the service, YOU ARE THE PRODUCT THAT IS BEING SOLD.
Take a ticket, and get some popcorn.
The mobile Internet market is busy defining 'unlimited data' as something else. More specifically;
- Throttling speeds after some arbitrary amount of data used. It really doesn't matter what the amount is, so long as they disclose it.
- Reducing video quality to reduce bandwidth demands. Which is a 'nice' way of saying your love of good quality video costs them too much, and they cannot afford to expand network capacity to satisfy your appetite for beauty.
- Working with video providers to cache the data closer to the mobile network, saving on peer network costs and enabling them to reduce video quality somewhat easier.
- Offering you higher data limits for higher fees, letting you set the price for what you want...
Why is this good?
> Throttling after a certain amount of data does let the user know they are profligate (by the definition of the seller) data consumers, and gives them the information needed to make buying decisions.
> Reducing video quality gives consumers a tangible measure of quality of service to make buying decisions with.
> When mobile and other ISPs develop partnerships with content providers they can better manage the data. Of course playing with video quality, for one, will expose to consumers the choices being made, and give them information to make buying decisions.
> Fractional data service may give some consumers more choice and control (Google Fi is a lot like this), though it may be that the oligarchy is merely letting consumers play with pennies.
But this is just the problem of supply and demand. Because it impacts our ability to watch kitten videos we go insane. Really.
deleting the extra space after periods so i can stay relevant, yeah.
You're right for a purely theoretical internet out of a textbook (though a very old one, using hubs rather than switches). A theoretical internet in which traffic on each link is consistent rather the peaky (no cable modems involved for sure), there is way more bandwidth available than is ever used, there;s no redundancy, meaning each packet takes the exact same path, and there are no buffers anywhere.
In the real world, we want our connections to be cheap, reliable, and fast. We want redundant, load balanced connections. Which means different packets take different paths - jitter even in a underused network. In the real world, we don't use just 10% of the available bandwidth, that would cost ten times as much. For lowest cost to users, ideally ISPs build links that aren not saturated 99% of the time, the 1% busiest times some will get saturated. Buffers are everywhere - every single connection point is a store-and-forward switch, not a hub.
>> Newflix, a new company wanting to compete with Netflix, might say "rather than our customers paying for a higher priced plan in order watch our service all night, we would like to pay the extra cost and subscribers with even the el-cheapo internet plan can use our service, because we'll pay the extra cost direct to Verizon."
> nothing in a net neutral world that stops Newflix from doing what you're suggesting. What the net neutral world is stopping is Newflix paying for that different tier to appear only if they are connecting to Newflix.
Why the hell would Newflix pay to upgrade my service for me to watch Youtube? Obviously they'd only want to subsidize the cost of using their service, not a competing service.
So, the solution is to burden users with choices? How is an uneducated person or a child going to shape her internet traffic? They have to learn that the internet is not just a big truck first.
Don't be fooled by this TREAT because it's actually a TRICK.
There is nothing to prevent an ISP from throttling back everything else to a higher extent than they are now, but will just use the excuse that YOU configured it. Technically, you already have the ability to control traffic via QOS or other means, but those are more complicated.
Opened the paper and kept reading and reading expecting at any moment for it to reveal what how it is supposed to work and finally just gave up. It was so loaded with this accomplishes x, y and z... while not being like a, b and c... that I gave up. In fact I did skim thru the rest but was unable to locate where that text was hidden if it exists at all.
Simple truth is there is as a political matter no possible workable QOS strategy across administrative domain on the scale of the Internet the same as there is no workable security strategy so I give up trying to decipher TFA.
What people CAN do is control egress from their OWN networks and intentionally drop incoming packets to manage ingress. Thanks to all of the snazzy congestion algorithms baked into TCP users have leverage to prioritize both outgoing and incoming flows by themselves. Consumer routers and popular third party firmware have had these features baked in for quite some time.
Where I strongly disagree with what little I have read is view CDNs are not a solution. ISPs used to run proxy servers that would accelerate ALL web traffic without special backroom deals back when cost of bandwidth was high enough to warrant but generally it sucked, often slowed things down and tended to cause a lot of breakage. My opinion with following minor tweaks this model can be made to work to the benefit of everyone. Operators, Users and Content.
1. Explicit browser configuration for upstream proxy servers where configuration is downloaded to provide browser managed load balancing/failover/sharding... This is opt in only. No transparent bullshit possible.
2. ALL content must have a current digital signature to keep ISPs honest.
3. Only content explicitly marked EXTERNALLY cacheable goes thru upstream cache everything else is bypassed.
4. Upstream has veto power to say FU go get it yourself to protect itself and keep content honest and ensure caches do not devolve into liabilities for ISPs. No I will not cache piecemeal 2k files, no I will not accept content with a lifetime of 30 seconds, nobody else is asking for this file so take a flying leap and get it yourself.
5. Cache only retrieves content when browser provides a current digitally signed proof of authorization by originating server. (No Legal traps/DMCA/redirect bullshit)
I don't know about the economics of massive ISPs my guess they may not be interested yet many midsized and small ISPs actually do invite CDNs into their networks just to save bandwidth operating on a purely mutually beneficial basis.
None of the above ? I'd like a steady stream with low latency at the bandwidth I was advertised, instead of the bait and switch the low down dirty scoundrels generally offer. During mid day and primetime TWC(Spectrum) fails to even resolve addresses reliably. Their DNS is configured on sequential IP's on the same subnet, forcing me to configure alternate DNS solutions e.g. Opendns, and Google DNS. My Amazon delivers a nice throughput but DirecTV on demand is jittery and pauses often.
errr....umm...*whooosh* *whoosh* Is this thing on ?
I don't think the folks at Stanford thought this through too much, or had the end-goal of killing Net Neutrality.
Twitter supports and protects racists - by smearing their critics with the "Hate Speech" label.
I think I'm clear on what you're saying now. Newflix could NOT offer free delivery, paying the extra cost of getting their service to you. (Unless they paid for everyone's high-speed connection to everywhere, which is a ridiculous proposition.)
I'm curious, do you want to make Amazon's free delivery illegal? How about Amazon Prime TWO DAY delivery - their packages arrive faster than other companies, because they pay to have it delivered faster. Should that be illegal?
If I buy a car and the dealer includes free oil changes for a year, they are paying for something that the customer would normally pay for. Do you think that should be illegal? What if they throw a gift card for $100 of free gas - they've bought the gas from Exxon and I get it free with the car.
These issues aren't cut and dry to me. I have X GBs included in my phone data plan. If Youtube wants to pay the extra cost for me, so I'm not charged for watching Youtube, I'm not sure that's a bad deal for me.
That sucks about the DNS. Good thing you have options available, options that are free.
Let me give a very specific example. Suppose you are sending five packets, which we'll call A B C D E. You've just sent A and B, which will each have 25ms latency. Packet C can be delivered in only 5ms. Do you want it delivered in 5ms, or should it be delayed, so it takes 25ms like A and B?
The right answer is "it depends". For gaming, you want low latency, deliver packet C as soon as possible. For downloads and Netflix, you don't care. For VoIP, you want packet C delayed; you want to make sure it's not delivered faster than the other packets.
Why do you want to slow down VoIP packets if needed to have consistent latency? Suppose those packets represent your voice actually saying the alphabet. If C is faster than An B, and D, the person on the other end will hear: ... D E. Faster packets change what you said.
C A B
That's one example, there are other cases when the three are incompatible, and many more where you you can gain alot of the one you care about by giving up a little of one you don't care about. So it's helpful to know what's important to you. That's even within a single stream - when I'm running both Netflix and ssh, I have more complex priorities. I want to reserve a very low bandwidth at low latency for the ssh, and a specific medium bandwidth for Netflix, but don't care about latency for Netflix. That may well mean that Netflix should be routed over the Cogent backbone and ssh should be routed over Level3.
Reading through this paper it seems all about providing a mechanism for ISPs to monetize content delivery preferences. It's hilarious that they throw in token statements like this:
Their proposal, Network Cookies, is all about identifying users and their content delivery preferences. Tell me how that doesn't hurt user privacy? Somebody above said this is like "Do-Not-Track." I disagree - this is more like an Uber-Mega-Persistent-Cookie which, according to their recommendations, will work not just in HTTP but all network protocols they can get their hooks into.
There are many inconsistent statements throughout this paper. At one point they label DPI witha "high transaction cost", at another it's "low overhead."
> Does that make sense?
Sure does. It seems our thinking isn't as far apart as I first thought.
> That's the fear. That someone will dominate the space super effectively.
I guess that's where we differ a bit. I think millions of pages of laws is already too much, so I'd prefer to err on the side of fewer laws rather than more. We're not too good at predicting the future, so I'd rather wait and see what happens and address things as needed, rather than being quite so proactive, making big laws based on the fear of what might happen. That also makes illegal a lot of good things that might have happened, and makes existing good things questionably legal (blocking spam would be illegal under one net neutrality proposal.)
EXACTLY! There has been enormous work on providing various levels of quality of service for streaming and other transport. To date, virtually none has been implemented in the Internet, and the net neutrality rules have, probably correctly, prevented content providers from imposing QOS differences on net traffic they provide, especially if it involves deals with carriers. But it should be possible for the consumers of content to decide they're willing to pay more to get better quality streaming services. Paying to avoid a digital hiccups during important airings should be an acceptable practice. All this assumes, of course, that the world's backbones are willing to evolve enough to support the QOS specs in the first place. Until now, the only successful approach has been overprovisioning, which admittedly works surprisingly well most of the time.
Most of those pages arose from specific examples of people being asshats. Also, I don't understand why people think the law can be simple. A fairly constrained problem, like an OS, takes 10's of millions of lines of code, plus documentation, etc.
It depends on how easy it is to undo. Breaking up monopolies is far hardr than preventing them, esp., since the monopoly has value that then has to be confiscated when broken up. But with some things (is this dun laoded) you fail safe, and with others (should Joe be able to start his custom shoemaking shop) you can fail dangerously.
I would consider running the internet to be a fail-safe situation.
It depends on what you mean by spam blocking. I don't see how it would be against any net neutrality operation. But if it did, it would only be one of like a million ways that spam is blocked, and the gains are huge.
Your ad here. Ask me how!
I'm in Midwest USA, 6 hops from Frankfurt Germany AWS and less than 2ms of jitter. I get microbursts from AWS Frankfurt. Back-to-back 1500byte packets hitting my firewall at 1Gb/s for 30-50ms at a time. My ISP is exactly the theoretical textbook network that you first described. I get my full 150/150 connection speed to nearly 100% of the datacenters around the world 24/7, all for the low low price of $50/m. Stop being an apologist for crappy ISPs.
>It depends on what you mean by spam blocking. I don't see how it would be against any net neutrality operation.
Net neutrality means traffic from all sources are treated equally. Emails from well-known farms are not treated differently than emails from your boss.
While reasonable people can make reasonable interpretations of a neutrality *policy*, there's little room for reasonableness when the *law* says "traffic from all sources must be treated equally". One NN bill introduced in the US house of representatives was about that simple - it was very clear, and no exception was made for spam, DOS attacks, etc.
MOST attempts to send email currently are spam which is blocked by an ISP of one kind or another. The spam you see is spam that got through, the 1% that wasn't blocked.
For instance, right now if you try to send email through the network of Germany's largest ISP (DTAG), to and from rtfa@gmail.com, the ISP won't allow that - and shouldn't. Only gmail servers should be sending email from @gmail.com.
> Certainly gmail can filter the e-mails you receive.
Google can prevent spam sent to your @gmail.com address, but Comcast has to deliver spam to your @comcast.com address? I guess we'll all be forced to use web mail.
> Right, the pipes aren't allowed to distinguish between the two types of e-mail. It'd be very hard to write a law that allowed that without breaking net neutrality
And if a law is passed that doesn't distinguish them WELL, it's going to create significant problems. As in the most-used network security measures will be affected (firewalls consider the source) , spam, routing optimizations, hell even basic protocols required for a network to function at all discriminate by source - STP to prevent switch loops, EIGRP, OSPF, and BGP would be required to accept poisoned routes (not allowed to trust some sources more than others). Maybe half of all network administration that's beyond "Networking 101" considers, in some way, the source of the traffic.
> But DDOS, pretty much by definition, is something an ISP cannot detect, let alone prevent. After all, it looks like a bunch of individually innocuous requests.
If DDOS attacks couldn't be detected and stopped, whitehouse.gov would be down ALL THE TIME. There's always somebody trying to attack it. 99.98% of the time the site is up because the defenders are winning.
They can be detected and prevented, and they are. You say "by definition", I believe you're thinking of one very specific type of DDOS, and also assuming that the defenders are pretty stupid. You're thinking of naive flooding attacks. Not all DDOS looks like legit traffic. Also, in the types in which a single request looks like it could be legit, that doesn't mean you can't notice a swarm of identical requests from known botnet. As an example of the first, an attack I invented would have easily taken down wikipedia and other major sites. It involved sending a DNS request from a source other than the source in the ip header. The ISP *closest* to the source could detect the attack and block it because it"knows which networks that endpoint has routes for. For example, a cable modem can't actually route traffic for other public networks. Therefore they could detect that it was a bogus packet. The recipient can't tell, but the ISP can, because the ISP knows it has no routes to the claimed IP.
For requests that otherwise appear legit, even floods can often be detected. A request for /signup.asp from a person using IE6 on Mac might be legit. 10 million simultaneous requests for /signup.asp from 100,000 people using IE6 on Mac? Not legit. Those of us who have been defending networks for 20 years aren't stupid. Well some of us aren't. Sometimes the bad guys get ahead for an hour or two, but you'll notice that for DDOS attacks the good guys almost always end up winning - we find a way to block them.
> Please provide more details, because I do not believe it works the way you do.
Then as I said, try it. Try making an SMTP (mail sending, TCP port 25) connection to any mail server, Gmail, Yahoo, or Slashdot.org, using your home internet connection. Every mail server obviously ACCEPTS mail at the server, but you probably won't be able to connect? Gmail and Yahoo aren't denying you, so why can't you make any connection on port 25? Because YOUR ISP knows that your connection is a consumer connection, not a mail server, and therefore shouldn't generally be sending mail directly. You must connect to their mail relay and probably authenticate. Again, if you don't believe me, try it. (Or read Networking for Dummies Part 2).
Try sending any packet at all with a source address of 1.2.3.4. You won't get anywhere because your ISP will block those packets. Your ISP, and only your ISP, knows that you aren't allowed to route for 100.2.3.4. Neither the recipient nor anyone else can possibly know whether you're permitted to route that network, so only your ISP can stop it. Seriously, if you don't believe me, try it.
You might also Google "egress filtering".
Submitter is a corporatist apologist. And WTF is this nonsense about letting consumers "choose" to have their bandwidth throttled. BeauHD must also think that banks are doing depositors a huge favor by paying a $1.25 overdraft in return for a $35 fee, plus $7 a day until the depositor gives them their blood money.