Slashdot Mirror
Who Pays for Rebuilding the Internet?
pcause writes "The Internet (physical as opposed to technical) was really not designed for applications that want to use maximum bandwidth all of the time, such as P2P and streaming video. Here in the US we've seen Comcast try to balance the demands of P2P traffic with other traffic and its backbone capacity. In the UK, a flame war has broken out between the BBC and ISPs about the same issue. So the question is who pays? Should the content owners who make the profits pay for the extra infrastructure, or should the consumer pay?"
The interesting question is not who pays, but how can we all collectively pay less for better performance? The problem is that the billing model for the internet is broken. ISPs need to start billing for usage, that much is obvious. But in addition, I think it would be really interesting if they billed based on a function of their actual cost for delivering every individual packet. I.e. if it stays in their network it's really cheap, if it goes through a peer then it's still pretty cheap, and if it goes to a transit provider then it gets expensive. the upstream ISP could in turn bill based on their cost to deliver it. Routers could pass along this metadata about the cost, accumulating it along each hop.
Obviously this has tremendous implications in terms of the additional work that routers would need to do to account for traffic, and how the costs are communicated to the customer. However, I think the end result would be something quite incredible, because what would happen is it would drive the development of smarter P2P protocols that keep traffic nearby, and widespread deployment of caches for static content and such. Right now there is very little incentive to do these things.
The end result, once everything has had a chance to adapt, would be a phenomenally efficient internet, with reduced costs and better performance all around. ISPs wouldn't give a hoot about this new class of "smart" P2P because the bulk of the traffic would stay among their local subscribers, the bandwidth to whom is free. Massive loads would be disappear from peering centers and long distance links. The cost of bandwidth would plummet.
I think all of this is feasible, and it's worth doing.
A better question would be, "why is the market broken?"
Well, here is the catch. Who ever spends the money should gain control of the resulting infrastructure. If the BBC/British government pays to upgrade the lines you can expect a great big (politely worded) fuck you to the telecoms if they try to set any demands.
If the telecoms pay for the infrastructure, they get to say what happens to it. Within whatever terms they negotiate for the use of public land to build on. And if they continue the false advertising of their services, they can expect that at some point a class action lawsuit will be made and will break them.
..adjust the rates based on time of day (or generally, demand at a given instant). There's a ton of spare bandwidth at night.
they both pay (consumer and content owner). They even pay according to the bandwidth they're provided, in most cases. Exactly who does the writer think is getting free service?
"National Security is the chief cause of national insecurity." - Celine's First Law
Consider all those folk choking up the internet with video downloads, P2P etc "because it's free" and they've got nothing better to do with their time. They're all choking up the pipes for everyone.
Pay-per-use is one way to get a free market into this and allow people to buy the QoS they want and the market decides the price points. It would also motivate some of the spam botters to clean up their act.
The only way to modify behaviour is to provide some sort of feedback/dis-incentive for excessive use.
Under pay-for-QoS you'd have the choice to wait til 2am and download a video for free or download it immediately for $2.
Engineering is the art of compromise.
For too long, the ISPs knew that their infrastructure would eventually get burdened down with data. It's not what kind of data, per se, but how many users are going to start requesting and sending data. The amount of users accessing the internet has reached a large enough point that the development of social networks seemed obvious. Every ISP knew about the potential to deliver video, but they must have underestimated the adoption of it by .....about ten years. How come we don't all have fiber to our houses? Better question, how come I don't have DSL and I live in a city of over 200K people? I had a cable modem since 1997 and all I can get is Comcast. When asking Qwest when I could get DSL, they insisted it would be in my neighborhood in two years. Two years go by and I actually called them. They said the same thing! Two years, sir. Eleven years have passed and I still have to be within a two mile radius of their main line or it's a no-go. If demand for faster speeds is so high, why, I ask is there no competition in my poor city?
Only now is Verizon taking charge, seemingly, to lead fiber directly into a single residence. I do remember a story in the late 90s about the former owner of Qwest saying he acquired the rights to 'improve' the length of most railroad lanes. Supposedly, he laid fiber in one tube and left the other empty so when the next, best-quality fiber became available, he could fill that tube. I don't know why I mentioned that, other than to illustrate the fact that people have been 'thinking' of upgrading the infrastructure, but it appears that they haven't.
If I had just made major investments in laying the major infrastructure for anything, I'd try to max-out my investment, too (aka, milk it for as long as you can before people get sour over quality.) Comcast has a maximum amount of bandwidth they can reasonably supply to a shared pool of users. When they try to balance users' internet demands while trying to sell more users on high-bandwidth HDTV and VOIP, they are just setting themselves up for having to do what they've been, rightfully, criticized for --throttle the most-demanding applications. While I strongly disagree with their actions, I guess I could understand it it I were a CEO wanting to impress shareholders, or if I were an IT manager wanting to get a raise.
In the end, I think we will see standard market forces doing what they do best --compete. I don't like the fact that I may have to give my money to Verizon, but I'd sure rather give it to them in the future (whenever that may be) than give it to Comcast.
No sig for you! Come back one year!
What we're looking at here is the Quality of Service question.
Different services require different service characteristics. For instance:
- File transfers can take "best effort" service. They don't care if the transit time of packets varies (jitter) or is long (large latency). They don't care if occasional packets get lost or corrupted because they can have them resent. They don't care if the rate is fast or slow - and can self-adjust to go as fast as possible to use the available bandwidth.
- Streaming protocols care about all of the above: If they're 2-way interactive they care about latency. They always care about jitter. They don't want packets to drop - but if occasional packets DO drop it's better for them to NOT try to get them resent, which would create massive jitter and latency. They have a bandwidth requirement that is either constant or related to what they are carrying - and has no relation to the actual speed available to the connection under varying amounts of congestion.
To serve both of these types of traffic on the same packet-switching network you have to treat them differently. Otherwise the file transfers will speed up to try to hog all the available bandwidth, dividing it evenly among themselves, and stomping on the streaming protocols. Things will only work for the streams on a best-effort net when all the file transfers are limited by some OTHER bottleneck and there is enough bandwidth ON EVERY HOP for essentially all the stream packets to go right through.
But because the streams have some other inherent bandwidth limit you CAN treat them differently. You can give them Quality of Service rules that puts them at the head of the line, limiting jitter, minimizing latency, and causing other types of packets to drop while they go through. And you can reserve bandwidth for them, refusing to set up the flow (connection) if there isn't enough to service them and have some left over for other services. Then you can guarantee they get delivered, while the file transfers etc. expand to hog and divide only the UNreserved bandwidth. Also: If they're going to multiple destinations you can use multicast and reserve bandwidth for only one copy while serving many endpoints.
And IPv4 was designed to do much of that: It has a "type of service" field that lets you declare what kind of service would be good for each packet. It didn't do bandwidth reservation (by itself). But you could declare preferences about latency, jitter, and drop probability.
Unfortunately, this means that packets could ask to be treated better than their competition. And it was completely on an honor system. And before streaming was widely deployed Microsoft deployed an IP stack that "improved" their product's performance by lying about the type of service the packets really needed, demanding stream-type service for everything, including file transfers. This got widely deployed. So ISPs generally don't honor the Type of Service bits, and QoS isn't widely deployed on the backbone.
Nowdays, in addition to the fast-as-mercury, dumb-as-rocks backbone routers, there are reliable-as-telecom, smart-as-firewalls edge routers, full of arrays of processors so they have a bunch of instruction executions available to think about every packet. These boxes can do things like act as a reverse-firewall to protect the network against cheaters, certifying that, if a given customer has bought - or temporarily reserved - a certain amount of high-QoS bandwidth he doesn't exceed it, and if necessary rewriting the QoS/Type of Service tagging so the backbone can trust the packets again.
So with the brains available to watch over the packets and apply rules, so that streams can get the bandwidth they need and file transfers can use all the rest on a common network, the question is what rules to apply.
Streams put a higher demand for service on the net - but have limited bandwidth. File transfers want bandwidth but are happy to take what's left after the strea
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
Bragging about the size of the BGP tables in your border and core routers just makes it seem like you missed the point. Frankly, I think you did, because the issues discussed in the article aren't about propagating the inter-host routes themselves, but about those hosts that already have routes using more and more bandwidth for things like streaming video.
If you have some personal knowledge that implies that BGP tables are in fact the limiting factor that's causing the friction between ISPs and content providers, then you should state that, and back it up. Some of the reasons that posts like your parent often talk about expanding the infrastructure with public money is because laying cable is hideously expensive and often requires other public support such as rights-of-way; these barriers to entry are why such infrastructure is often viewed as a "natural monopoly". We haven't became the top of the food chain in this planet because of some fag-democratic ideal. Cruelty and ruthlessness made the Humankind the Kings of this world! Democracy is for the weak and the losers... Wow. I may be falling for a troll here, but really . . .
What operating system do you use, personally?
Microsoft Windows? Your network stack was originally implemented in BSD-derived code, but (if you use Vista) was finally replaced.
Apple's Mac OS X? Your network stack is implemented in BSD-licensed code.
Any Linux distribution? Your network stack is implemented in GPL-licensed code.
So, really, it doesn't matter what operating system you use, it wouldn't have been able to get on the Internet without the people whose "fag-democratic ideals" you disparage. That's true for all of the people going through your border and core routers, as well. Think about that the next time you feel like bragging about the size of your, um, BGP tables.
I'm surprised that no-one has picked up on this yet... as far as I know, this is the first time that a major TV channel has actually made its content available online for free. Although it is a good example of how whiny the ISPs are, there is another point - HD, full-length videos of a TV series are now freely and legally available to the public. This by itself demonstrates that the infrastructure is going to need a major overhaul over the next few years.
Most human behaviour can be explained in terms of identity.
As a consumer and not a content provider, I believe the consumer should pay. I should purchase a certain level of service. The ISP should tell me what level of service they offer, I should make my decision, then they should provide that level of service. And the penalties for anti-trust violations and monopoly violations should be astronomical.
Why? Because I want to be the person that controls the purse strings. I want to be the person deciding what level of service is appropriate. The last thing I want is for a few dozen major players to make that decision without my direct input.
The person closest to the purse strings makes the decisions. That means that if you want to make the decisions, you have to be the person closest to the purse strings. You want to be the person who is getting charged for QoS. That gives you the power to decide what QoS should be. The last thing I want (and the last thing I think any of us want) is for Internet service to work the way cell phones do.
Please, charge me. Give me the power of the purse. In an amoral capitalist economy, the power of the purse is the only power that matters. Unless we're figuring on going socialist, I want to be the decider.
Stop-Prism.org: Opt Out of Surveillance
Having a monopoly on the infrastructure won't solve the problem. In fact, that's pretty much the very definition of the problem. What incentive would this private company have to sell access to the fibre to other ISPs at a reasonable price?
We pretty much have this situation in Australia, as Telecom Australia (as a taxpayer funded government department) built and owns all the telephone exchanges around the country, backhaul capacity between them and the cities, and so forth. All the ISPs have to buy backhaul capacity from Telecom (now "Telstra", a non-government, listed company), beg them for access to exchanges in order to install their own DSLAMs, and so on.
Granted, it doesn't help that Telstra are also in the retail market directly competing with ISPs, but even if they weren't -- what incentive do they have to keep prices at commercially fair and competitive levels? They don't. There is no competition, so they can charge whatever the hell they want. If someone does try to compete, they can then lower prices in that small region, which means no other company is going to invest in infrastructure to compete with Telstra because they'll never make money on it.
Here, the government heavily regulates Telstra and generally does all the things that governments should never do to businesses. This is a mess and it is clearly not a long-term solution, because we (the taxpayers who paid for the infrastructure in the first place) are getting screwed anyway, and it's also incredibly unfair to Telstra.
"I don't recall anyone ever saying "To have a free market, it must be provided by public Government services", a free market can never have any Government regulation or intervention, else it is not a free market."
Hogwash and wordplay. There is no free market without regulations and governments to impose them. An unregulated market is not free, it's a playground for the strongest party to create a monopoly.
If you don't recall anyone telling you this, maybe you should make it a point to educate yourself on the subject of economics before you start spouting your opinion. All important thinkers on economics understand that regulation is necessary in order for a market to be free and fair.
---
"The chances of a demonic possession spreading are remote -- relax."
Who paid for it to begin with?
If you are located in the UK, why not try out the UK Free Software Network ISP? All their profits go to open source software funding, they set well defined badnwidth limits and good speeds, and don't interfere with your network traffic like some ISPs do. I haven't used them before, but it looks awesome and I will definately be switching in the near future.
This is how the loudness war is killing music.
This is such a ridiculous argument I literally cannot believe the ISPs are trying it on like this! Providing bandwidth to consumer is what the ISPs do! People are using the ISPs product more, surely that should be a good thing for the ISPs!? No, because they fucked up the industry too much with 'unlimited' deals. Well I say "Fuck 'em", they can sort it out. Free broadband is obviously not a viable option so why did they start it? The ISPs will have to start charging more for broadband to pay for the increase in demand. That's how the market works, why the fuck should BBC have to pay a dime!?
Which sounds pretty, but that same short life means the cost of purchasing drops quickly too.
I've worked in networking at an ISP myself, and if you're comfortable riding the back of the wave, you save a lot of money and can still keep up with bandwidth needs.
Trouble is, the company has to keep upgrading all the time, and I'm sure that Bell, etc, operate much like my company did - the managers resent the constant costs and see sparkles as they count the potential huge savings by delaying purchase/rollout by 6 months.
It's a constant trade-off: Unhappy clients vs Costs to keep up
The top managers *love* arguments (comes with being evil) - my director used to cut bandwidth payments and pretend to be hard up, just to force negotiations with the other company. Sucked for us techs, but he loved the fights. And the company's would usually cut a deal, saving our company $30k-$100k at a time. Keep flipping this around, and it really adds up.