• Featured Articles

Comcast/Level3: Should Other CDNs Care?

December 6th, 2010 by · 4 Comments

Over the weekend I read this piece by StreamZilla’s Stef van der Ziel taking Comcast’s side of last week’s public dispute with Level 3, and it got me thinking about just how it affects them.  The obvious temptation is for pure CDNs see Level 3 as a competitor who uses its ownership of a tier 1 backbone to lower its costs and thus gain the advantage of a lower cost structure.  And that Level 3 seeks to do this is not in question, owning the network is a key ingredient of their whole approach.  But I think that pure CDNs also have something to fear from a last mile provider bending transit networks over a barrel.  Why?  Because the rates that major CDNs pay to companies like Comcast for paid peering are intertwined with the fortunes of the IP transit marketplace.

CDNs prefer to peer directly with larger last mile providers due to performance, however what if the price were too high?  They can always pay for transit from a tier 1 transit network for the same traffic in a pinch.  In fact CDNs all pay for transit to reach endpoints they don’t have direct agreements with.  If they do that, then the last mile provider gets nothing – hence their willingness to price their paid peering for CDNs competitively.  But what happens if transit networks find themselves limited by traffic ratio, such that they can’t deliver substantial amounts of video content without paying extra fees?  One can imagine negotiations like this:

ISP-X: In order to better match our cost structure, we are doubling our pricing for paid peering.

CDN-Y: That’s too high!  You will force me to buy transit from someone like Cogent or Global Crossing instead.

ISP-X: You’re welcome to try, but as soon as their traffic ratios rise then they pay us the same increase. We expect they’ll pass that on to you somehow, but we don’t really care.  We get paid either way.

CDN-Y: But I can’t get to your customers any other way.

ISP-X: Not my problem.  You could always build a last mile network like we did… [chuckle]

CDN-Y: But what if you double pricing again next year?

ISP-X: Hmmm, that’s not a bad idea.  Thanks!

This has come up now because Level 3 explicitly owns and operates a CDN, and won a big contract – thus provoking an early conflict.  But as video increasingly dominates bits delivered to consumers by all networks, other transit networks will face increasing traffic ratios with providers like Comcast whether they operate their own CDN capabilities or not.  That’s why Global Crossing is also wary of the possible precedent.

The IP transit business is cutthroat, a fact which has kept prices falling steeply and steadily for a decade.  It’s a business that is not much fun, and one that CDNs like Akamai have made a great living out of bypassing.  But a bypass is a bypass only in relation to the original route.  Separate the economics of video delivery from the economics of transit and you risk losing the competitive underpinnings that keep the pricing for delivery of bits falling steadily.

It is tempting to blame all this on Level 3′s insistence on running a CDN and leveraging its Tier-1 backbone status, but it’s all about the traffic and the future of traffic is video.  Do CDNs really want to say that carriers should pay last mile providers if they carry too much video?  Because if that happens, it’s not just the carriers’ costs that will rise.

That doesn’t necessarily mean other CDNs should endorse Level 3 in this conflict.  I’m just saying that it’s a bit more complex than ‘Level 3 is undercutting us and should pay more’.  One must be careful, lest one get what one wishes for.

Categories: Content Distribution · Internet Backbones · Internet Traffic

Join the Discussion!

4 Comments So Far


  • Poppa says:

    Question. The CDN delivers a video, but isn’t the last mile customer the one who makes the demand? If there is increased cost associated with the increase in demand, wouldn’t it make sense to charge the consumer of that demand, not the provider? If I want more potato chips, the grocer doesn’t charge Frito Lay more money for the product. Why should the end provider charge the CDN for proving the wavy chips?

    • Avatar of Rob Powell Rob Powell says:

      I tend to not try to argue the “whose fault is the traffic” line, because it is circular. Both sides benefit, and each end of the pipe thinks it is the one adding to and the other subtracting from the value.

      But if the grocer were the only one available in town, and would lose popularity with his own customers if he raised rates – I’ll bet he would try to get more money out of Frito-Lay…

      And that in and of itself is not my concern. I’m just worried that competition will be lost, and innovation stifled along with it. My nightmare is that the FCC comes in and regulates traffic exchange with the last mile carriers. The name would be “special access peering”. *shudder*

  • Eric says:

    The whole transit imbalance argument is slightly ridiculous.

    It makes sense if you are talking about largest networks like level3, sprint, cogent. In that case, peering occurs fairly early in the routing process, so a traffic imbalance means that one provider ends up carrying the traffic further and using up more of their capacity. Traffic ratios matter because you don’t want to be carrying another provider’s traffic all around the country/world (a number of US providers rely on others for Asia for example).

    However, a last-mile provider does not care about imbalances–users download much more content than they upload. The lower traffic ratios between Comcast and Level3 earlier indicate that level3 is not being used by content companies like youtube and netflix for the last mile connection.

  • Stephan says:

    Actually, your Fritos / supermarket analogy is apt. Most of the big supermarket chains have a great deal of leverage over consumer product and food producers, since they’re the primary distributors of branded foods in the US. The big chains charge “slotting fees” of food producers in order to prominently display their products on store shelves. See http://en.wikipedia.org/wiki/Slotting_fee or http://www.wisegeek.com/what-is-a-slotting-fee.htm.

    Unsurprisingly, this practice makes it more difficult for new or small brands to get their products on the shelves of major chains.

Leave a Comment

You may Log In to post a comment, or fill in the form to post anonymously.





  • Ramblings’ Jobs

    Post a Job - Just $99/30days
    Jobs by Simply Hired
  • Event Calendar