Interesting People mailing list archives
more on the circus of public peering
From: David Farber <dave () farber net>
Date: Thu, 20 Oct 2005 12:07:47 -0400
Begin forwarded message: From: "Patrick W. Gilmore" <patrick () ianai net> Date: October 20, 2005 10:58:15 AM EDT To: dave () farber net Cc: "Patrick W. Gilmore" <patrick () ianai net> Subject: Re: [IP] the circus of public peering Dave,I would note that InterNAP is not doing well financially, has never made a profit, is not and has never been an important network on the Internet like Sprint, Cogent, France Telecom, etc. This is not to say InterNAP is irrelevant, just that their size, scale, traffic, etc., are really not comparable to those of the "big" networks.
Saying "InterNAP did it this way" is the same as saying "my mom's fruit stand worked like this (until she filed for bankruptcy), so I don't see why the large grocery stores can't do it."
-- TTFN, patrickP.S. My work has some rules regarding posting to public forums, this is more for your personal info. If you feel necessary to post this to IP, please remove my name.
On Oct 20, 2005, at 9:32 AM, David Farber wrote:
Begin forwarded message: From: Ophir Ronen <ophir () singlestep com> Date: October 20, 2005 12:36:52 AM EDT To: dave () farber net Subject: the circus of public peering Dave,While I've been out of the backbone business for some time now, I do have a modicum of insight into this problem. My former company, Internap, was created to resolve this specific issue. Internap built an infrastructure of PNAPs (private network access points) each of which connected to the ten largest Internet backbones; this provided excellentcoverage of the publicly reachable IP address space.Internap rapidly became a very large customer of these backbones because each new PNAP necessitated a new circuit to each of those backbones. Each PNAP facility acted as a broker between the Internap customers attached to the PNAP and the backbone to which their traffic was intended. Internap customers compensated Internap and Internap subsequently compensated each backbone for their share of the traffic.This situation created excellent economies of scale in the pricing of these interconnect circuits as well as providing a series of high performing and financially viable peering points. The backbones were compensated for the traffic sent to them and, in most cases, also enjoyed lessened pressure on their public peering points. This was due to Internap sending them direct customer traffic rather than transittraffic.IMO, the notion of public peering is flawed and has been since the Internet became commercial.disclaimer: I have not worked for Internap for >4 years and have very little interest in them. I'm simply describing a model that worked without relying on public peering.cheers. -Ophir Ophir Ronen CTO Singlestep Technologies Connect. Simplify. Manage. P: 206.448.9700 F: 206.448.2666 E: ophir () singlestep com https://www.linkedin.com/profileviewProfile=&key=88179The information contained in this transmission is privileged and/or confidential information intended for the use of the individual or entity named above. If the reader of this message is not the intended recipient, you are hereby notified that any use, distribution or copying of this communication is strictly prohibited.On Oct 19, 2005, at 1:23 PM, David J. Farber wrote:===== Forwarded message from Daniel Berninger <dan () tier1research com> =====\From: Daniel Berninger <dan () tier1research com> To: dave () farber netSubject: Level 3's de-peering of Cogent threatens the inter in InternetDate: Wed, 19 Oct 2005 18:26:57 -0400 Dave,For IP if of interest. The dispute between Level 3 and Cogent remains unresolved with Level 3 planning to de-peer Cogent again on November 9,2005. Best regards, Dan ............................................ Daniel Berninger VP, Senior Analyst Tier1 Research v: 202.250.3838 e: dan () tier1research com w: www.tier1research.comLevel 3's de-peering of Cogent threatens the inter in InternetThe anti-competitive instinct that leaves the US a laggard in Internet access arrived as a threat to the Internet backbone business with Level 3's de-peering of Cogent on Wednesday, October 5, 2005. Level 3's move against Cogent marks an escalation of the tensions that threaten settlement free peering. It has the hallmarks of an anti-competitive strategy rather thanjust a business decision about the merits of peering. Businesses cancompete by improving the value proposition offered customers or they can attempt to injure the ability of other businesses to make competing offers. A wannabe monopolist accepts short term self injury as the price of reducing competition and winning future monopoly profits. Level 3's decision to make at least 15% of the Internet unreachable did not improve the valueproposition it offers customers.A world without settlement free peering is a world without an Internet. Level 3's test of the restraint of trade option comes only six months before SBC and Verizon come together with collective scale five times larger than Level 3 and with far more experience in allocating markets. The refusal to interconnect has always represented the weapon of choice for companies seeking reduce competition in the communication business. AT&T used refusalto interconnect to establish a monopoly over telephone service at thebeginning of the last century as did the Bell cartel at the beginning of this century. The antitrust remedy that led to the breakup of AT&T in 1984 arose to win interconnection for competing long distance companies. The story offered in Level 3's press release regarding Cogent comes straight out of the monopolists handbook. Level 3 disconnected Cogent to protect its"investment" and end the "free ride" enjoyed by Cogent.The increasing consolidation of Internet backbones inevitably brings out the toll collecting instincts, but the move against Cogent was less about a sizeand more about a disparity in business models. Level 3 CEO Croweconsistently laments the lack of pricing power during quarterly earnings calls. He seems to blame Cogent for this lack of pricing power, but he should blame himself for debt and lavish spending on a network that prevent him from remaining price competitive. Punishing Cogent for keeping transit prices low does not begin to solve Level 3's troubles. Level 3 pays out in interest 25% of every dollar in communication revenue, but the company would still have a negative net income without interest payments. Level 3 enjoys an enterprise value equivalent to 75% of what Verizon plans to pay for MCI, but MCI remains 10 times larger than Level 3 in communication revenue. Level 3's problem remains the same as other backbones in the Internet access bottlenecks that limit traffic growth. Success in raising prices transitprices will only reduce traffic volumes further. The motivation for settlement free peering decreases as the disparitybetween networks increases, but a concern for the long term health of the Internet should remain a part of the decision making process. Level 3's toll collecting ambitions conflict with its embrace of companies like Skype as the source of traffic to fill its network. The promise of the Internet depends on continuous cost performance improvements like every other area of the info tech industry. Level 3 benefits from cost performance improvements in processing power, storage, Internet access, and everywhere else in the IT value chain. Any success Level 3 achieves in raising prices will throttlethe Internet's growth. Rising costs will shrink the types of viableapplications and associated data traffic setting in motion a vicious cycle where backbone companies will need to raise their prices still further. The replacement of settlement free peering with a toll collecting ethic means aless connected and lower performing Internet.Level 3's plan to raise backbone transport prices can't succeed by just de-peering Cogent. Level 3 needs SBC (AT&T) and Verizon (MCI) to de-peer Cogent. Qwest and Savvis (acquired C&W) look like the next most likely candidates to lose their Tier 1 ISP credentials. Level 3 might not even make the cut if SBC and Verizon decide to embrace Sprint. The smaller backbones can use settlement free peering between each other to minimize dependence on the Tier 1 cartel, but Level 3's move against Cogent leads inevitably to a period of arrested development. The US government has proven a friend of market power in the battles over Internet access. Acartel of US companies controlling the Internet backbone provides yetanother reason for the UN to assert a role in Internet governance. If Level 3 proceeds with the plan of de-peering again on November 9, 2005, they should revise the empty sentiment of their "the network partner you can relyon" motto. ===== End forwarded message ===== ------------------------------------- You are subscribed as ophir () speakeasy net To manage your subscription, go to http://v2.listbox.com/member/?listname=ipArchives at: http://www.interesting-people.org/archives/ interesting-people/------------------------------------- You are subscribed as patrick () ianai net To manage your subscription, go to http://v2.listbox.com/member/?listname=ipArchives at: http://www.interesting-people.org/archives/interesting- people/
------------------------------------- You are subscribed as lists-ip () insecure org To manage your subscription, go to http://v2.listbox.com/member/?listname=ip Archives at: http://www.interesting-people.org/archives/interesting-people/
Current thread:
- more on the circus of public peering David Farber (Oct 20)