In a blog post yesterday, Level 3’s Mark Taylor laid out the case that today’s peering connections is under attack by the last mile. He didn’t name names, but he also did it without invoking that content company that begins with the letter N that has so dominated the discussions lately and focused in on the current state of affairs in the company’s peering relationships.
Basically, of Level 3’s 51 peers, only six are both congested and not in the process of upgrading. Those unnamed but rather guessable six providers have been letting the situation persist for over a year. All six are major consumer last mile operators, with 5 in the US and one in Europe. No coincidences there, or surprises either.
The implication, which was stated by Stacey Higginbotham over at GigaOm more explicitly than Level 3 dared, is that these providers are deliberately abusing their market power. Thus in order to get better performance, content providers like the one that begins with ‘N’ will feel forced to pay the last mile directly for throughput that nobody else is allowed to provide. It’s not as if upgrade foot-dragging is a new tactic in the world of interconnection, but if the pattern of who is doing it to whom is the same with other backbone operators it ought to raise a few eyebrows.
I’ve been amused by some portrayals of the peering/transit interconnection layer lately that make it seem like a highly competitive Disney-esque environment where everybody is happy and has dozens of equally viable choices they are inexplicably not making. Level 3’s partial revelations are at least a start toward the transparency we need in order to understand what’s really going on and judge what is abuse and what isn’t. Might other providers step up to the plate and divulge a few details? I expect we’ll get an earful from Cogent’s Dave Schaeffer on their earnings call this week, but will anybody else speak up with a few such details?
For now though, does anybody want to guess who the six unnamed consumer broadband networks are? I would be shocked if Verizon and AT&T aren’t on the list, and probably CenturyLink too. That leaves probably two cable MSOs out of the usual suspects (Comcast, TW Cable, Cox, and Charter). I’d pick the two biggest but Level 3 did come to terms with Comcast just 10 months ago — would they be congested again already? In Europe it’s probably Orange I’m guessing, but the game is different over there so I figure the focus will be on the 5 on this side of the Atlantic.
If you haven't already, please take our Reader Survey! Just 3 questions to help us better understand who is reading Telecom Ramblings so we can serve you better!
Categories: Internet Backbones · Internet Traffic
Join Our LinkedIn Group
The difficult peer in Europe is likely Deutsche Telekom (DTAG).
DTAG has for the past 5+ years restricted ports/capacity to peers and has a US sales team selling paid peering to ISPs and content.
Pricing for DTAG German paid peering is typically 6x the price of full route transit.
what do they do when you buy full transit but only send on-net traffic over the port? sextuple the bill? :p
Eyeballs have been valued more than content for at least a decade. As far back as the early 2000’s I can recall providers offering free pull traffic in order to help balance their peering traffic ratios.
I think the best way to solve the problem is more competition in the last mile. Municipal fiber (if the municipality has the technical competence) or conduit banks make economic sense. Dropping a 48-strand SMF cable in a trench when a new water or sewer main is being installed has a marginal cost so low it can almost be ignored.
When content providers pay Level3 to deliver content, Level3 calls it Enterprise transit revenue and deems it just and good.
When content providers pay AT&T, Verizon, CenturyLink, Comcast, or Time Warner to deliver content, Level3 calls it extortion and suggests government intervention to save the Internet from the evil robber barons.
Level3 argues they have to haul more bits further.
ILEC/MSO companies argue they have to haul more bits to more places.
Both are legitimate perspectives. This comes down to who gets hurt more by congested peering: the consumer or the business who relies on having their traffic delivered around the globe.
Re: calling it extortion, I think there is no problem Netflix or anyone else paying the last mile for content delivery as long as it isn’t artificially forced. The problems arise if congestion is used to force that by deliberately degrading the alternatives through such neglect. Without that, there are no problems here.
Frankly, the likes of Netflix probably *should* be buying paid peering if they want their own CDN infrastructure and a quality experience. Transit shouldn’t be able to offer the same performance as that just given the additional distance and hops required to reach the eyeballs. But if it isn’t even a viable option, there isn’t a market.
Why should Netflix pay for something that an ISPs customers have already paid for? How could that ever NOT be artificially forced? Comcast accepts money from subscribers every month to provide access to the Internet, all of it. It’s then incumbent on Comcast to provide that service, not provide as much as is convenient and limit access to content they think they might squeeze cash out of.
It always has been a powerplay. For many years L(3) has used it’s market power to force people into transit relationships. Now the cable companies and other broadband players are finally waking up and they had some success in negotiating with folks like L(3).
I strongly believe that it is a combination of both the Powerplay that leads either to congested ports or to settlement based solutions leading to a crippled Internet.
The endusers should pay their fair share for the network costs they generate. The Transit providers should be paid by their customers. CDN folks should be paid by the content owners. And everything in between should be Peering.
It helps keep traffic local, scales pretty well and is for the good of the Internet.