The Internet Transit market and the gravitational pull to $0/Mbps.
last updated August 2010
Internet Transit is a simple way to get connected to the Internet.
Click here for "Definition of Internet Transit"
You simply pay a metered rate for your traffic to be delivered to the Internet and for your network to be announced as reachable across the Internet. Since the commercialization of the Internet, Internet Transit prices have declined every year as this article will discuss.
In this report we document the U.S. Internet Transit market's twelve year quest towards $0/Mbps pricing.
Since 1998, DrPeering has been tracking the commodity U.S. Internet Transit pricing.
The method for data collection is informal, mostly based on discussions at Internet Operations Forums from 1998 to 2010. It is important to note a few things about these informal surveys:
Every year, everyone believed that the Internet Transit pricing drops could not possibly continue. And yet the gravitational forces pulling transit prices downward appears to be a natural law.
"No one can make money at $___ /Mbps" and "the pricing has to level off now" they said. Yet every year the pricing dropped again. As you will see in the data below, it is not a good time to be in the Internet Transit market as a supplier.
Internet Transit Prices (1998-2014) U.S. Internet RegionYearInternet Transit Prices (in Mbps, min commit) % Decline
per Mbps 1999
per Mbps33%Source: DrPeering.net
The unmistakable Transit Pricing trend is down, with an average decline of 61% from 1998 to 2010 as shown in the graph below.
U.S Internet Region
Content Delivery Network pricing is following the same downward price trend, and tends to be priced at approximately the same price point as Internet Transit. One implication of this is that the CDNs that don't run their own backbones (such as Akamai) may fare better than those who do run their own backbone (Limelight Networks) since Akamai is able to purchase transit at increasing volumes and at ever decreasing costs. (Transport prices tend to drop at a lower rate.)
Dan Rayburn shared CDN pricing information on CDNs that showed pretty amazingly low pricing trajectories ($0.02-$0.06 per GB for 500TB commits) approximately in line with our estimates.
So here we take a stab a projecting the future - predictions of this type are merely for discussion purposes.
The Transit and CDN Markets are doomed. The days of simple Internet Transit or vanilla CDN services are numbered; and any business model that is based on margins on something going to zero are doomed.
The Peering Sector will morph in two ways to serve the large (video scale) peers. Since Internet Peering is an alternative to simply sending your traffic through Internet Transit services, the question, as always, is "Does Peering Make Sense Anymore?" The answer will be YES, but for a different market than today.
With Internet Transit Pricing dropping so fast, peering using Public Peering ports will become very difficult to justify financially, and the volume of traffic between bilateral peers will exceed the amount that is reasonable to do over public public ports.
Click here for "Discussion about the Cost of Internet Peering"
To illustrate, let's do some simple calculations and see how this plays out given our transit pricing history and predictions data.
If you assume the cost of public peering is $10,000/month ($2500 for a 10G public peering port and $7500 for colo, power, transport, equipment, etc.), you see that in 2010, peering makes sense financially if you can peer away at least 2Gbps (2000Mbps) for free to your peers. This gives you plenty of breathing room to peer traffic away for less than the price of transit, up to 7.6Gbps (7600Mbps), which is approximately the amount of traffic one can reasonably sustain over a 10G peering port. But this is for 2010 with an average transit price of $5/Mbps.
overhead associated with ethernet framing, and some buffer to not drop packets during bursty periods.
See the sidebar discussing the cost of peering.
YearTransit Price (per Mbps)Peering Cost (per month)Peering Break Even PointEffective Peering Bandwidth (w/10G port)2008$12.00 $10,000 833 Mbpsup to 7600Mbps2009$9.00 $10,000 1111 Mbpsup to 7600Mbps2010$5.00 $10,000 2000 Mbpsup to 7600Mbps2011$3.25 $10,000 3077 Mbpsup to 7600Mbps2012$2.34 $10,000 4274 Mbpsup to 7600Mbps2013$1.57 $10,000 6369 Mbpsup to 7600Mbps2014$0.94 $10,000 10638 Mbpsup to 7600Mbps2015$0.63 $10,000 15873 Mbpsup to 7600Mbps
With these pricing estimates we see that in 2012, peering still makes sense, but with an average Internet Transit price of $2.34/Mbps, we see a Peering Break Even point of 4274Mbps where peering begins to make sense. Beyond this point, peering makes sense financially up to 7.6Gbps, the Effective Peering Bandwidth of the pipe.
Definition: Effective Peering Bandwidth refers to the maximum usable bandwidth available for peering traffic, taking into account framing and burstability. As peers approach the Effective Peering Bandwidth, they will generally need to upgrade the interconnect.
This is not much of an Effective Peering Range; getting peering traffic volumes to fit in this range is tricky, so peering folks will mostly likely not go forward with public peering (all things remaining equal).
Definition: Effective Peering Range refers to the range of traffic exchange from the Peering Break Even Point to the Effective Peering Bandwidth along the peering path. This is the range of traffic exchange where peering provably makes sense financially.
As fewer and fewer ISPs publicly peer, the rationale for public peering to aggregate smaller peering sessions together over a public peering fabric diminishes. And the technology is not well suited today to handle massive peering sessions over public peering fabrics.
As a result, we will make the prediction that Private Peering will dominate the peering ecosystem in the U.S. by 2012. Point-to-point fiber is just too cost effective, reliable, and ease to deploy to compete with.
Definition: Private Peering is Internet Peering across transport with exactly two parties connected to it, usually a fiber cross connect or point to point circuit.
Most Internet Exchange Points offer or facilitate the interconnect of two parties using some form of dedicated point to point interconnect known as cross connects or circuits in addition to offering a Public Peering fabric.
The sector will morph to serve primarily three classes of IX customer over the next few years:
There is simply no margin in Internet Transit services. There had better be some form of premium or bundled services to offer on top of transit services to provide some profit margins. ISPs then act like a convenience store that offers milk below cost so that customers buy the other items in the store while you are there.
Never before has there been such an opportunity for the Access Networks (those who connect the eyeballs to the Internet) to charge a little more to content companies for enhanced access to the eyeballs. Paid Peering for content providers, embedding services for content, selling services within the last mile, are all examples of premium access services that could be sold to content who will pay for them. If you are a CDN and your competitors are embedded in the last mile networks, you pretty much have no choice but to purchase the same level of access just to compete.
A wise man named Hank told me "The Internet industry pathology is that dollars are not flowing through the supply chain."
In order to deal with the loss of margin on the transit services, ISPs may join forces in the form of cooperative initiatives to enable the sale of premium Internet services via an overlay network supported by the InterStream, ETICs, etc. or other industry-specific cooperative VPN services such as the old ANX (Automotive Network Exchange) promote. The power in these initiatives come from the unique services provided for customers from these cooperative network services. Our work with large carriers suggests a 30% premium could be realized solely by offering end-to-end cross AS boundary services.
With a cross-AS cooperative CDN service for example, all ISPs across a supply chain may reap the portion of the incremental revenue that large video streamers are willing to pay for a better end-user experience. This would disseminate the CDN revenue across those who distribute the bits, not just those who peer those bits away from a disk at a peering point to an access network. This is one example where ISPs who cooperate in such end-to-end services can offer services at a premium that other ISPs simply can not provide.
The author was invited as part of a team brought together by a large Venture Capital firm in Silicon Valley. The challenge on the table -
"Design a business model for a CDN that prospers when the price for distribution $0/Mbps".
It was an interesting exercise, and the conclusions we reached were that CDN revenue will be derived from new sources:
1) The value is not in the bits, it is in the data surrounding the consumption. Consider a new Hollywood sitcom that does a preliminary test run in an Internet test market before they go into full production. When do consumers turn off the show - how far in before they give up on the pilot? Refining the formula and running again is possible with different groups within a test market. Do we have the winning formula for the show? That is critical information to the content producer, one that increases the chances of their success and, yes, something they are willing to pay for. What is missing is the business model and the technology.
2) The value will be in the add-ons - getting the bits there is not enough - some indication of the reliability, some guarantee for the quality of the streams delivered, having instrumentation and control panels that meet the enhanced needs of these large video pushers continues to determine who wins the business. The services that are built on top of the commodity transit will be the ones to charge for - transit is the milk that is priced low to get people into the convenience mart.
3) Taking a piece of the value - cooperative and revenue participation based service offerings will allow the content distributed at $0Mbps and pay out when the clicks, purchases, etc. are made.