Let’s get our priorities straight

by on August 17, 2010 · 26 comments

Adam recently pointed to Robert Litan and Hal Singer’s new Harvard Business Review essay in which they defend ISPs’ right to offer special handling for certain packets. They write, “The ability to purchase priority delivery from ISPs would spur innovation among businesses, large and small. Priority delivery would enable certain real-time applications to operate free of jitter and generally perform at higher levels.”

I’ve recently begun to see this sort of argument used in the net neutrality debate more frequently. Where once the mantra of the opponents of regulation was that, thanks to competition, there has never been a serious case of discrimination so that government intervention was unnecessary, now we’re hearing more and more that innovation and investment will suffer if ISPs are not allowed to offer priority services over the internet.

What Litan and Singer explain about price discrimination and product differentiation is absolutely correct. What I don’t think they understand is that “priority delivery” of content over the internet is not within ISPs’ technical capability.

First, let’s be clear what we’re talking about here. As far as I understand it, net neutrality regulation would apply to communications over the internet, not simply to communications that happen to use internet protocol (IP). Consider services like ESPN360, or edge-caching a la Akamai, or imagine Google colocating their YouTube servers at Verizon’s offices. These techniques allow users to get packets over the internet quicker, but do nothing to violate neutrality. You can also imagine Comcast offering an IP-based video game service that, much like cable TV, creates a fast direct connection between Comcast and the user. In a sense, this is prioritization of packets, but not over the internet. If I’m not wildly off-base, this would not violate any of the major net neutrality proposals, either.

So if we’re not talking about colocating, edge-caching, or creating separate IP-based networks, how exactly would ISPs offer “priority delivery” of packets over the internet? The very nature of the internet is that it is a best-efforts network. Short of a reengineering of the net, any “additional or differentiated services,” as outlined in the Google-Verizon proposal, would have to necessarily stand apart from the internet because, as much as they’d like to, ISPs cannot prioritize packets over the internet. If they could, we’d see such a service now, but we don’t.

Why can’t they do it? Because it would require the dozens, if not hundreds, of networks that a packet traverses in its travels from sender to recipient to agree to respect the same prioritization scheme. Even if we assumed that we could reengineer the internet to accomplish this, you’d have to deal with users disguising low-priority packets as high-priority ones. And because a central arbiter would likely be necessary, you’d probably lose some of the unplanned nature of the internet that makes it so wonderful.

If that’s the case, then short of reclassification of broadband as Title II, net neutrality regulation that prohibits prioritization of packets over the internet would be regulating something that ISPs can’t technically do. (Again, I reiterate here that I’m assuming that edge-caching, colocating, and separate IP-based networks would not be prohibited by neutrality rules.) So I’m starting to ask myself what’s so bad about the Google-Verizon proposal, especially if you exclude the wireless sector, where there is competition.

I’m not saying I support the deal; I’m still very suspicious of it. But I’d be curious to hear what folks think about my thesis that what it does is give the FCC power to police crimes that will never take place: prioritization over the internet (which is not technically possible) and blocking (which has been adequately checked by competition thus far).

  • http://bennett.com/blog Richard Bennett

    Two points: 1) Jerry says: “The very nature of the internet is that it is a best-efforts network.” I see this a lot, but it doesn't make any sense at all. The Internet Protocol deals with packet routing, not with packet delivery. In its entire life, IP has never delivered a single packet. All it does is find a route for a packet, and then place the packet into a queue for delivery by a Link Layer function. The Link Layer (and its underlying Physical Layer function) actually deliver the packet, so any decision about whether to go Best Effort, Better than Best Effort, or Cheaper than Best Effort is fundamentally a Link Layer decision. The nature of Link Layer protocols is out of scope for Internet engineering, following E2E or not. Link Layers are things like Ethernet, Wi-Fi, 3G, Wi-Max, and that sort of thing. They're all capable of various priorities by design. IP has the ability to request priority treatment via Diff Serv and MPLS. The Internet is not a “best efforts network,” it's an internetwork that can use whatever capabilities its underlying networks provide.

    2) Preserving priority end-to-end isn't that complicated. Transit networks are paid by edge networks to provide services based on the edge network's requirements. In some cases today the agreements that bind transit networks require them to honor priority markings. This is enforceable because the transit networks are not out there moving packets for the fun of it, they're businesses doing what they've been paid to do. The fundamental requirement for end-to-end priorities is an agreement between edge networks to respect priority markings where they peer (interconnect.) This can be accomplished by establishing quotas for high-priority packets. For example, the peering agreement between AT&T and Comcast might say that no more than 10% of packets can be marked high priority. It then becomes incumbent on each carrier to ensure that they abide by the agreement, which may mean “demoting” high priority packets for users who have exceeded their quota of same. Alternately, the receiving network demotes packets that are outside the agreed-upon limits.

    To make this happen, transit networks play a vital role. They can facilitate priority peering by advertising the fact that their customer networks support it. It then becomes possible for networks who don't have direct peering agreements to participate in a prioritization program. This is already happening in a limited way today.

  • http://jerrybrito.com Jerry Brito

    Thanks for your comments, Richard.

    Could it be I'm using “best efforts” colloquially and not in a specific engineering sense? That is, it would be quite a shock to me if you told me that the popularly understood nature of how the internet works–that it does not guarantee delivery, that there is no guarantee of quality of service, that it is the applications at the end that check for and correct errors, that it is generally dumb in the middle–is not in fact the way it works. Am I completely wrong about these general assumptions?

    On your second point, my question is this: Are we seeing priority service offerings now? If not, why not?

  • http://twitter.com/binarybits Timothy Lee

    The Internet is not a “best efforts network,” it's an internetwork that can use whatever capabilities its underlying networks provide.

    And the service actually provided on the real Internet is almost always best effort. It's theoretically possible to imagine a tiered Internet, and it's not hard to sketch out how such a network would work from a purely technical perspective. In fact, people have been doing just that for more than a decade. Yet there's been little if any movement toward offering such services on the real Internet. If this is such a great idea and so technically feasible, why hasn't anyone done it?

  • http://bennett.com/blog Richard Bennett

    Disqus has discarded my last comment a couple of times, let's see if this one sticks.

    With the rise of the CDN, the Internet is not really a “best efforts” system at the transport level, if it ever was. CDN's put their packets at the head of the ISPs' delivery queues relative to packets entering the ISPs' edges at the hot potato routing points closest to the origin server.

    Some people argue that CDNs don't violate the best efforts assumption, but that's disingenuous at best. Why do companies like Netflix pay companies like Akamai for use of their CDN if they don't improve performance? And how can they possibly improve performance except at the expense of services that don't use CDNs?

  • http://twitter.com/binarybits Timothy Lee

    How can they possibly improve performance except at the expense of services that don't use CDNs?

    If the bandwidth bottleneck is upstream from the CDN server, then the CDN would be using bandwidth that would otherwise be sitting idle.

  • http://twitter.com/binarybits Timothy Lee

    Also, Jerry specifically said that edge caching/CDN isn't what he's talking about, and I think most NN advocates would say that CDN's wouldn't be banned under their preferred regime. Now, maybe you think that distinction is irrational, but CDNs aren't a counter-example to Jerry's argument.

  • http://bennett.com/blog Richard Bennett

    Yes, in that scenario they would, but most people agree that the bandwidth bottlenecks are more likely to be in the last mile. But that's orthogonal to the issue: you claimed there are no tiered services in the Internet – a claim you've been making for a while now – but I just pointed you to one that's in very wide use.

    Are you claiming that CDNs aren't tiered services, or do you want to define tiered (or premium) services so narrowly that we can ignore the one that's right in our faces?

  • http://bennett.com/blog Richard Bennett

    Jerry's confused about how The Internet works. He says: “[CDNs] allow users to get packets over *the internet* quicker, but do nothing to violate neutrality.” His assumption is that packets are routed from origin network to some cloud called *the Internet* and then to the destination network. Jonathan Zittrain and others use this model as well.

    Sorry, but it doesn't work that way. Packets move from the origin network to whatever transit network the origin network is paying to whatever transit network the destination network is paying and then to the destination network. When the origin and destination networks are large ISPs such as AT&T, Verizon, Comcast, or Global Crossing, that means they flow directly from the origin network to its nearest connection with the destination network, and therefore the destination network pays for most of the delivery cost.

    There's no neutral cloud in there, everyone's getting paid for what they do.

    Consequently, the CDN jumps the destination network's queue and interconnects closer to the end user. So it's not really reasonable to discuss premium delivery service after ruling the prime example out of bounds. It's like saying “Let's talk about racial discrimination in the USA, but there will be no mention of black people.”

  • http://bennett.com/blog Richard Bennett

    In other words, this is wrong: “…as much as they’d like to, ISPs cannot prioritize packets over the internet.”

    All that ISPs have to do to prioritize packets is modify their bilateral agreements with other ISPs such that prioritization is part of the agreement. The Internet is nothing more than a set of bilateral agreements, and whether edge-to-edge behavior is tiered or flat is simply a matter of what the agreements say.

    The difficulty of changing these agreements is much less than the difficulty of changing the Internet from IPv4 to IPv6, for example, and the IP transition is already happening. There is a resistance to change in any large system, but with sufficient will, it will always be overcome when there's a good reason. Look at the cell phone network, 8 times larger than the Internet by number of devices, yet it goes to a new generation of technology and protocols every 3 years.

  • http://twitter.com/binarybits Timothy Lee

    It sounds like you're just raising a semantic objection to the way I (and I think Jerry) am defining “tiered services,” which is about routing policies rather than network architecture. Having a faster or better-placed point of connection with a destination ISP isn't a “tiered service” as I'm defining the term, and I think it's probably not a tiered service as leading NN advocates define the term. Maybe that definition doesn't make sense, and it certainly would be hard to codify into law (Which is one reason I think we shouldn't try) but I don't think it's an unreasonable distinction to draw conceptually.

  • http://twitter.com/binarybits Timothy Lee

    The fact that you don't like Jerry's definition of neutrality isn't evidence that Jerry doesn't understand how the Internet works.

  • http://bennett.com/blog Richard Bennett

    I showed why I think Jerry's description of the Internet was wrong, Tim:

    “There's no neutral cloud in there, everyone's getting paid for what they do.”

    Is that unclear?

  • http://bennett.com/blog Richard Bennett

    I'm showing that most NN advocates are inconsistent, permitting one type of premium service and forbidding another on no rational basis whatever. That approach makes for bad public policy and disadvantages communication-oriented apps at the expense of content-oriented apps. The NN people are trying to turn the Internet into cable TV, and I want it to be a picture phone that does cable TV in its spare time.

  • http://srynas.blogspot.com/ Steve R.

    You have touched on a very important subtlety to the net neutrality debate. “The very nature of the internet is that it is a best-efforts network. Short of a reengineering of the net, any “additional or differentiated services,” as outlined in the Google-Verizon proposal, would have to necessarily stand apart from the internet because, as much as they’d like to, ISPs cannot prioritize packets over the internet. If they could, we’d see such a service now, but we don’t.” Given this, you are correct to say that regulation would be difficult. But is also quite true that the lack of regulation would not magically guarantee improved service either.

    Just prior to reading your post I read the response of Dave K to a New York Times article. Dave wrote:“The other aspect to the “competition” argument, which a computer scientist should well know, is that the network connections in question aren't just me and my ISP, but also every ISP who owns any router or line in between me and whatever other machine I'm trying to connect to (e.g. a Google web server).

    Let's say I'm downloading a page from nytimes.com. I connect to my ISP, who is net neutral and allows me a nice 50 Mb/s connection to nytimes.com. However, nytimes.com is connected to another ISP, who isn't net neutral and has been paid big bucks to favor traffic to washpost.com over traffic to nytimes.com, so as a result nytimes.com can only give me a 50 b/s connection. Since this system always takes the worst connection speed, I'll get my page at 50 b/s despite not having any choices in this matter. I can switch to a competing ISP all I like, and that won't change one bit.”

    Despite all my lambasting of the Times, they have a good discussion here: Who Gets Priority on the Web?

    I reiterate, that other posts that I have read speak in terms of the US being behind other nations. So the question is not simply regulation versus no regulation, but what are these other countries doing that makes them better?

  • http://bennett.com/blog Richard Bennett

    Priority service is not about web sites, Steve, it's about new services that need something from the ISP that web sites don't need. The technical standards are already in place to provide this, and all it takes is some business arrangements to turn it on.

    There is not really a cloud of independent ISPs in the middle carrying packets willy-nilly, there is only the small number of transit networks who were paid by either the origin network or the destination network to carry packets for them on their terms. There is no neutrality in the Internet, you only get what you pay for.

  • http://bennett.com/blog Richard Bennett

    The description of the Internet as a best-efforts system is murky at best. Yes, TCP does error recovery, serialization, and rate control. But there is enough intelligence underneath the Internet these days (in MPLS and optical Ethernet) to ensure high reliability and to allow traffic shaping. The Internet is actually more like an application than a network. The people who run around talking about stupid networks don't understand the technologies that carry IP datagrams, do the routing, and cross the oceans.

    Priority services are available to commercial users, and are widely used. CDN's are the best type of prioritization for the home user, but other kinds are coming.

  • Pingback: A Question of Priorities | High Tech Forum

  • http://twitter.com/binarybits Timothy Lee

    I think your paraphrase of Jerry's description is a straw man. He didn't say there was a “neutral cloud in there” or deny that people are getting paid for what they do.

  • http://bennett.com/blog Richard Bennett

    This exchange has inspired a blog post at High Tech Forum: http://www.hightechforum.org/a-question-of-prio

  • http://bennett.com/blog Richard Bennett

    He said a packet traverses “dozens, if not hundreds, of networks …in its travels from sender to recipient.”

    Nope, doesn't work that way.

  • Larry

    I hesitate to jump in late into a technical conversation, especially to play Devil's advocate. But isn't the concern over “priority” “premium” or “fast lane” service that worries the Net Neutrality advocates limited to first and last mile, particularly the latter, over which the ISP does have control if not technical ability to moderate the flow of packets?

  • http://bennett.com/blog Richard Bennett

    Primarily, but not exclusively. The whole idea of good service for non-web apps scares the neutralists, because they see the decline of the web as something akin to the death of virtue.

  • Larry

    Right, I understand the separate/related concern about managed services or what the Google-Verizon proposal called “differentiated service.” But I guess my point was simply that it is technically possible for ISPs to discriminate for or against “traditional” web content on the last mile.

  • http://bennett.com/blog Richard Bennett

    Yes, it always has been possible for ISPs to block or degrade any kind of traffic they want inside their networks, and they use this power every single day to block DoS attacks.

  • Bvannice2003

    I just saw this thread.

    Richard i am a little confused about some of the things that you have said. You state that priority is a link layer function.

    It certainly can be but i don't think any service providers use the link layer mechanisms. In fact it makes the problems discussed in this post harder not easier. If IP networks were completely homogenous, built with Ethernet, for instance, it might make sense to use a (singular) link layer prioritization mechanism (802.1p).

    But they aren't. So if a provider chose that strategy they would have to intermedite priority among all the differnet links layers within their own network (Ether, DS3, SONET, ATM etc). And then figure out how to intermediate those same things with others. It would be very hard to do.

    You also refer to transit arrangements, which exist, but peering is more common. And in peering both sides agree to terms of service. And both have strong incentives to adhere – to keep their own customers happy. They also have incentives to differentiate their services – as in all markets. Prioritization provides that. If it weren't for the silly debate about Net Neut it would already be happening.

    I agree with your overall premise that priorization of traffic is achievable – but it will happen at the IP Layer. I can not see link layer mechanisms playing a role.

  • Marivic

    i know this site is not the right to publish or search a people,but i ddnt'know what iam doing right now.i search mary ann delrosario hall in 1070 san miguel rd.,concord C.A i never had any communication about her since i was born,she never know about our my mother died.hoping anybody see this add.have a good heart pls be inform me.,0920-8620389,0919-2530659,473-3395,EMAIL ADD:marivicledpalma@yahoo.com….PLS DONT BE IGNORE THIS MY ADD.THANK YOU VERY MUCH…

Previous post:

Next post: