Following up on a previous discussion, a demand for “net neutrality” usually means a demand that the network must not discriminate between applications being carried in IP packets; that identical transmission characteristics (throughput, delay, number of errors, etc.) are to be provided for all packets regardless of what is being carried in them.
But a demand for “net neutrality” is usually also wrapped together with a demand by these same people for no metering, no usage charges. This would mean that users who are continuously transmitting and receiving packets would pay the same flat rate as someone who is paying only for a typical traffic profile.
If this principle were applied to electricity, it would be like having no electricity meter. Everyone pays the same, regardless how much power they use. The problem: if you’re one of the 99% of normal users, you would have to pay DOUBLE what you normally would, to cover the costs of the 1% of users constantly drawing 200 amps 24 hours a day, 7 days a week, 365 days a year.
Here’s how that would work:
You, along with the other 99% of people in your city use maybe 50 amps, at various times during the day, most of the time a lot less… a usage profile that fits within a pretty narrow bell curve. Knowing this usage profile, how much money it costs to fire up those oil-fired generators to produce that much electricity, and the costs to develop capital infrastructure to deliver it, we can calculate what the monthly flat rate should be.
But hold on, it turns out that we haven’t taken into account the net-neutrality advocates, who are the 1% of the people in your city who would take advantage of this to draw the max: 200 amps 24 hours a day, 7 days a week, 365 days a year.
To pay for them, the flat rate has to be raised. Everyone has to pay more.
OK, you’re thinking, raised a little more, what’s the big deal? So everyone pays, what, $53 per month instead of $50 to cover the 1% of full-bore users who are drawing 200 amps 24/7/365 ?
Here’s the problem: it ain’t just a bit extra. The power used by the 1% is equal to what you and the other 99% use together. The 1% double the demand. So the price increase to pay for this 1% of users is closer to $50 in the example above. Instead of paying $50, which is your fare share, we would be talking about everyone else paying more like $100 to cover those 1% of people who use a LOT more power than you do.
Sound like a good deal?
Yes, if you are one of the 1%, (the net-neutrality advocates!)
… but not if you are one of the 99% and use pretty much the same amount of power as your neighbors.
And the net-neutrality advocates make this sound like it is a noble objective?
They have the moral high ground?
Carriers are “bad” and “corporate” because they give these 1% a choice:
– pay the same and use the same, or
– pay more and use more,
instead of making the 99% subsidize them?
Give me a break.
And don’t forget what they are doing is basically stealing… the disingenuous part.
ALL “NET NEUTRALITY” ARTICLES:
Net Neutrality – Foolish, ignorant or disingenuous?
Net Neutrality II: If the power company allowed this, your electrical bill would double
Is the Internet a Public Utility?
“NET NEUTRALITY” AND IP NETWORKING TECHNICAL RESOURCES:
TERACOM VIDEOTUTORIAL V9: QoS and Traffic Conditioning:
http://www.teracomtraining.net/previews/TERACOM-VIDEOTUTORIAL_V9-qos-and-traffic-conditioning.wmv
http://www.youtube.com/watch?v=dSNhcJDuA2s
Course 101 Telecom, Datacom and Networking for Non-Engineers,
Chapter 3-3 “Bandwidth-On-Demand: Packet Network Services” and following
Course 110 Understanding IP Telecom: IP, VoIP and MPLS for Non-Engineers,
Chapter 14. Quality of Service in the IP World: NET NON-NEUTRALITY
Video Course DVD-4 “Understanding Networking 1”
Part 3 WANs – Bandwidth On Demand: Packet Network Services
Telecom 101 Textbook
Chapter 18 Bandwidth on Demand
Hi Eric,
Ok, I’ll bite. To use your example, what the telco’s are proposing is to only give me 2 amps for my dryer and 200 mA for my stereo, when I am paying for 50 amps metered. It shouldn’t matter the devices I use, they should meter my total consumption irregardless of what I use it for. It is a utility.
What they are actually doing is illegally looking into my house (spying on me), seeing I use it for a dryer, and reduce the current going to that device inside my house.
If you want to meter by time of day or limit the wattage I consume, fine. I’ll sign a contract for that. But you cannot limit the number of amps going to a specific device just because you can.
Thanks for your comment, Peter.
I agree that if you have a network services contract that limits the total power you can use during a time frame, this goes a long way to solving the problem. It causes the user to restrain themselves, i.e. police their own traffic, or pay extra for overage, effectively putting them into the “use more and pay more” category.
But that doesn’t mean that we don’t need prioritization at network nodes. Networks are built based on statistical multiplexing, or overbooking. When more packets are transmitted than the statistics used for designing the network, there is congestion. This has to be resolved somehow… and we are going to have to resolve the congestion in favour of applications that are delay- and loss-sensitive (live voice and live video) and to the detriment of applications that are file transfers (web browsing, email, legal music downloads, illegal content-stealing etc.).
However, the big problem that has brought this to the forefront is persons that have a contract that does not specify any caps on the total amount used, a so-called ‘unlimited’ plan. 1% of users take advantage of that to be in the “use a LOT more and pay the same and everyone else can subsidize me/us” category.
I have to take exception to the statement that the Internet is a utility.
It ain’t. The Internet is a business.
And network service is not and has never been all-you-can-eat-at-access-line-speed flat-rate service. It can’t be, as networks are designed based on statistical multiplexing. The capacity internal to the network is less than the total of the access line speeds. If the network was designed for everyone doing all-you-can-eat-at-access-line-speed, the internal capacity would have to be increased by an order of magnitude. Network service would cost 10 times a much.
Network service is a Service Level Agreement, a contract between the user (traffic profile) and the network (transmission characteristics). I prepared a video tutorial explaining the idea of a SLA: http://www.youtube.com/watch?v=hTzk_oOrGrU
Now, if you want to set up an IP network and operate it as a utility, you are certainly free to do so. But you will very quickly find out that you have to *pay* to connect your IP network to other IP networks so that your subscribers can transmit and receive packets to machines not on your network.
Then you will find out that there are two basic kinds of interconnect services for sale on the free market: peering and transit. Peering is the cheaper option, and doesn’t usually include termination guarantees, the network you give your packets to will likely pass them on to one of their peers to get rid of them as quickly as possible. Transit is more expensive, as it usually includes termination and delivery guarantees.
Then you will find out that no-one offers unlimited plans. Only Service Level Agreements based on the RFC2475 Differentiated Services (http://www.faqs.org/rfcs/rfc2475.html) and implemented using MPLS, with transmission characteristics like RFC3246 Expedited Forwarding and RFC2597 Assured Forwarding are for sale.
So you, as the Utility IP network, do some research to determine, based on statistics, what service levels you need to buy for interconnect. You pay for those, then try to break even (if you are a not-for-profit utility) or make some money (if you are a business) charging retail for subscriber access, and providing transit and peering services to other IP networks.
And being altruistic, you make your Utility IP network service unlimited and application-agnostic.
Then 1% of your users run bittorrent 24/7, and it turns out that they use as much interconnect bandwidth as the other 99% of your customers.
So you have to double the interconnect bandwidth, doubling your cost. But you can’t raise prices much. You are now in a money-losing situation.
Now, more and more bittorrenters sign up for your service since it is unlimited and application-agnostic. What do you do?
– Go out of business (many ISPs did that)
– Put usage caps on new subscriptions. (Nasty bad press for doing that, but your solution)
OK, now what about all the existing subscribers who have grandfathered unlimited plans? And all of the existing peering and transit contracts you signed on the sell side that have unlimited plans? Including universities?
You analyze the network traffic, and discover that the problem is 1% of your users are generating 50% of the traffic, using one or two applications. And these applications are being used for stealing copyrighted works without paying the authors. Your legal department warns you that on top of the network congestion, there is a risk that the copyright holders will sue you, the network provider, for allowing people to steal their stuff.
What do you do?
– Lose money until you go out of business
– Can’t put usage caps on grandfathered subscriptions.
– Ignore the legal risk
– Implement traffic engineering techniques in the network to police the 1% of users so that their overage is only transmitted when there is spare capacity, to ensure that
a) the 99% of users get what they are paying for and
b) you don’t have to increase your network infrastructure and its costs by an order of magnitude to scale the network to meet the demand of the 1%ers.
– or maybe plead with the bittorrenters to be good network citizens and kindly stop running that application, as your network was not designed to handle the traffic it generates
– other suggestions?
– – –
Last point: the network is not invading your privacy. It is not looking inside your house to determine what application is being carried in the packets.
An ingress device in the network is identifying flows of packets, examining the protocol headers in the first few packets in the flow, drilling down as necessary, until it has a confidence level what the application protocol being carried in the packet is. Then it marks the packet with a label (the L in MPLS).
This label, along with all of the others that identify that application, is associated in the routers with a Diff-Serv Behaviour Aggregate. The routers are programmed to provide designed transmission characteristics to each such Behaviour Aggregate, i.e. application, in the network.
So the network is not spying on you. It is not looking at the content, not looking to see what kind of file is being transferred, not looking to see who is the author of the copyrighted work the 1%er is stealing, whether it’s Bambi or bestiality or a zip file with the entire Zappa library or what.
It is just looking at protocol headers to identify the file transfer protocol: bittorrent, limewire, FTP, HTTP or SMTP etc. It is no more an invasion of privacy than routers examining other protocol fields in the packet, like for example, the destination IP address, to determine what label to apply.
This is done on a machine in a building owned by the network service provider, not in your home.
Thank you for your posts. I have been trying for some time to get a clear picture as to how a network works (I am uneducated and don’t wish to spend a semester at college taking a networking class when my degree program doesn’t require it). I think, based on your articles, that the networks should stay the way they are. I am a happy Hulu watcher, who pays more then a non-hulu watcher for faster service on ACS (based in Alaska). Our tiered system works well- no usage cap, users pay extra for faster connections, and everyone goes home happy. I must also point out that in my community, pirating is *huge*! Probably 40-60% of the users with my ISP download illegially, because there’s not much else to do. Talking casualy with agents of ACS, I have heard that their network is doing quite well… Personal experiance has shown very little congestion, and no downtime to speak of.
I do not think that users should be charged per-byte or what have you. It would cause people to use the net a lot less. A system where one pays for their speed in teirs with unlimited data, I am all for.
Unfortunately, the net-neutrality advocates have in their camp people who are against tiered systems, and people who are against usage caps.