DrPeering -

Why hasn’t anyone brought up Comcast Paid Peering and network neutrality?  Isn’t Paid Peering extortion?  As in... Tony Soprano saying “You want access to my customers? You pay me.”  <queue Sopranos theme music>

Sy Musico

----

Sy -

DrPeering sees two sides to this argument, and you are describing one side. Kudos for the imagery. 

The Peering Community seems to have (at least) two views.

Here is one view: Comcast Paid Peering is extortion because

1. 1)A CDN is paid to get content topologically close to the eyeballs.
   

2. 2)Akamai already has its equipment embedded inside Comcast’s network [Some debate/heat has arisen on this point].
   

3. 3)For a CDN to compete with Akamai effectively (on performance) its needs to get its content close to the eyeballs.
   

4. 4)Comcast charges a metered rate in exchange for getting content topologically close to the eyeballs (aka Paid Peering).
   

5. 5)There is no alternative provider that can be used to get closer to Comcast eyeballs.
   

There are even rumblings that performance of one of Comcast’s transit providers is sometimes congested, which makes Comcast paid peering even more attractive; paid peering is a lower-cost higher-performance alternative to traversing that transit provider to reach Comcast eyeballs.

But you have to pay for this high quality access to Comcast eyeballs.

How is this not two tiers of access?; a sometimes congested transit path, or a direct paid peering path that one network operator described as having twice the performance of transit. If you were a CDN, you would really have no choice.

[ Editor’s note: Time Warner Cable and other providers also have paid peering products. Several Tier 1 ISPs and AOL have had paid peering products for years and have marketed them to ISPs as stepping stones to free peering, and to content heavy ISPs as providing improved access to the Tier 1 ISP’s customer base. ]

Here is another view: Comcast Paid Peering is an enabler, a competitive threat to Google.

Google wants Paid Peering to be illegal, along with any form of access tiering.

This argument is draped in Internet freedom, let innovation thrive language, and Comcast made mistakes with the whole DPI and throttling p2p traffic thing.  But let’s put the religion aside for a second.

If you were Google, strategically, why would you want paid peering illegal ?

Killing Paid Peering will increase Google competitors’ cost, and decrease Google competitors’ performance. Consider the cost of traffic exchange graphic below.

Any company can buy Transit (metered access to the global Internet) for around $2-$9/Mbps (Source: 2009 Field data).  Google is paying about $0.50/Mbps for transit today (Source: RampRate).

To optimize performance and decrease costs, a company can siphon off the traffic destined to Comcast eyeballs and buy Paid Peering (direct access to Comcast network) for around $1-$3/Mbps.

Paid Peering enables Google’s competitors to get access to Comcast eyeballs for around the same price as transit ($1-$3 instead of $2-$9/Mbps).

Paid Peering provides better performance than Internet Transit, since the traffic takes a less circuitous route.  As shown graphically in the movie below, direct Internet Peering can provide 2-3 times better performance than the Internet Transit alternative.

Therefore, Paid Peering allows Google competitors to more easily compete with Google on performance and price without having to reach Google scale.

What happens when Paid Peering is made illegal?

Killing off paid peering would force content heavy companies, ISPs and CDNs to pay a higher cost to reach the eyeballs and remove for all the performance benefits of paid peering.

[Editor’s Note: These graphics point to Google in the <$1/Mbps range but there are others who also do the volume of traffic to get into this transit price range.]

To finish up this discussion, consider the hypothetical network diagram below.  Here we see several networks and CDNs that could be used to access Comcast eyeballs, each with different proximity values to the eyeballs.

Let’s explore a hypothetical example to highlight proximity choices facing content companies and see where paid peering might come into play.

In this hypothetical scenario, NetFlix could purchase from a transit provider like Sprint, but its content would be two network hops away from the eyeballs that the content guys want to serve.  Purchasing from a Limelight in this scenario would help push the content a network closer to the eyeballs compared with transit, but purchasing from a CDN that peers with Comcast (BitGravity in this example) would get the content even closer. Akamai is rumored to have nodes embedded within Comcast which would be the best (topologically closest) option if true. If NetFlix saw it as strategically important to run their own peering, they could perhaps seek paid peering with Comcast. 

One thing is clear -- if you are a CDN, you need to buy paid peering from Comcast to compete on performance with Akamai.

Summary -  Paid peering appears to be similar to a two tiered access network, one of the very things that the FCC may seek to regulate as discussed in section 106 of the discussion draft. 

We should explore next the negative consequences of regulation, some unforeseen negative consequences, and what the peers say are some of the benefits paid peering offers over unpaid peering. 

Dr Peering

DrPeering@DrPeering.net