5 ms·

The economics of the Internet backbone (2004) [pdf]

44 points by vmarsy 10y ago

walrus01 10y ago
Anything written about this in 2004 is going to be woefully out of date, with huge growth in traffic from CDNs and things like Netflix, YouTube and major companies like google establishing pops near large IXes to peer. Mid sized ASN here, traffic patterns and ratio of paid transit to peering is very different these days. Also transport and transit costs are dramatically different. Don't rely on anything this PDF says to get an idea of how ISPs handle peering, transit and layer 2 exchanges in 2016/2017. This article is so old it talks about MAEs... Jeez.
- NKCSS 10y ago
  Would be interesting to see if there has been any recent info on this, maybe you're aware of some?
- alex_hitchins 10y ago
  Agreed. If there was an up-to-date version of this document I'd be really interested in reading it. I am continuously fascinated with the infrastructure etc. of the internet.
  - walrus01 10y ago
    To a certain extent the people who have a really firm grasp on the layer 1 through 3 topologies of the Internet, as it relates to business, contractual, transit, peering, IX, CDN, L2 inter-city transport relationships between ISPs don't want to write a comprehensive guide to this... If you have a mid sized ISP and the persons who make the final contractual/purchasing decisions on transport circuits, colo space, router equipment purchases have comprehensive knowldge of this, and detailed knowledge/GIS data for the OSI layer 1 network topology of the geographical area in which they operate, it provides a huge competitive advantage over other competing ISPs in the same area. If you look at CAIDA's ASRank tool, some of those top-100 ranked ASNs in North America leveraged their senior neteng and management's knowledge to make shrewd business decisions regarding how/where to build their network. Conversely the less successful ISPs waste capex and network engineering time and resources extending their network in ways that don't have the same ROI.
- vmarsy 10y ago
  I found the paper very instructive. Could you explain what's so different today compared to 2004? The paper specifically says ISP can do 100% paid transit, or 100% peering , or anything in between. If the actual ratio transit/peering is different today, how does it invalidate the paper? The paper raised some good points that can be useful when discussing net neutrality, and also "Facebook Basics" kind of offers.
jnbiche 10y ago
> The Economics of the Internet Backbone > [by] Nicholas Economides What a singularly great name for an economist.
- grkvlt 10y ago
  Nominative determinism in action? See https://en.wikipedia.org/wiki/Nominative_determinism https://en.wikipedia.org/wiki/Nominative_determinism ;)

44 points by vmarsy 10y ago

Anything written about this in 2004 is going to be woefully out of date, with huge growth in traffic from CDNs and things like Netflix, YouTube and major companies like google establishing pops near large IXes to peer. Mid sized ASN here, traffic patterns and ratio of paid transit to peering is very different these days. Also transport and transit costs are dramatically different. Don't rely on anything this PDF says to get an idea of how ISPs handle peering, transit and layer 2 exchanges in 2016/2017. This article is so old it talks about MAEs... Jeez.

NKCSS 10y ago
Would be interesting to see if there has been any recent info on this, maybe you're aware of some?
alex_hitchins 10y ago
Agreed. If there was an up-to-date version of this document I'd be really interested in reading it. I am continuously fascinated with the infrastructure etc. of the internet.
- walrus01 10y ago
  To a certain extent the people who have a really firm grasp on the layer 1 through 3 topologies of the Internet, as it relates to business, contractual, transit, peering, IX, CDN, L2 inter-city transport relationships between ISPs don't want to write a comprehensive guide to this... If you have a mid sized ISP and the persons who make the final contractual/purchasing decisions on transport circuits, colo space, router equipment purchases have comprehensive knowldge of this, and detailed knowledge/GIS data for the OSI layer 1 network topology of the geographical area in which they operate, it provides a huge competitive advantage over other competing ISPs in the same area. If you look at CAIDA's ASRank tool, some of those top-100 ranked ASNs in North America leveraged their senior neteng and management's knowledge to make shrewd business decisions regarding how/where to build their network. Conversely the less successful ISPs waste capex and network engineering time and resources extending their network in ways that don't have the same ROI.
vmarsy 10y ago
I found the paper very instructive. Could you explain what's so different today compared to 2004? The paper specifically says ISP can do 100% paid transit, or 100% peering , or anything in between. If the actual ratio transit/peering is different today, how does it invalidate the paper? The paper raised some good points that can be useful when discussing net neutrality, and also "Facebook Basics" kind of offers.

jnbiche 10y ago

> The Economics of the Internet Backbone > [by] Nicholas Economides What a singularly great name for an economist.

grkvlt 10y ago
Nominative determinism in action? See https://en.wikipedia.org/wiki/Nominative_determinism https://en.wikipedia.org/wiki/Nominative_determinism ;)