The Internet Metrics post in December was a good starting point for my research on backbone traffic, but I'm struggling to get more specific information. Specifically, I am trying to find out the market share for the major backbone providers (so far I've found WorldCom 37% and Sprint 16% of traffic). ... Interdiciplinary Telecommunications Program, CU Boulder

i submit that your metrics are subject to serious question. while we expect little of boarwatch, i would hope academics would be a bit more responsible. e.g. how do you assign the traffic share of the traffic going to a smallish network which is multi-homed behind sprint, uunet, and which is also appears at mae-west? hint: if you look from a uunet or sprint perspective, then you assign it to uunet or sprint depending on perspective. but if you also peer openly at mae-west, then you don't assign it to either. so how about some prudent or accurate, as opposed to merely fast, ideas? randy

it is impossible. well, depending on what you are measuring, and how concerned you are with the accuracy. no accurate yardstick exists for such a comparative measurement. what you want to measure (traffic, ip addresses, web sites, ASes advertised, capacity, utilization, routers, or any other single network-based measurement) isn't even collected, counted, recorded, or archived in a consistent, similar, or even correct manner across all of the set of entities that you would likely desire to compare. ...all of which even assumes that you could /obtain/ that data (stats, routing tables, etc.) and required support from the set of entities that you want to compare-- which you couldn't. one could interpolate, extrapolate, estimate, and assume a whole lot to fill in blanks with the inaccurate and incomplete data that is available publicly, but it would likely result in yet another [perhaps well-meaning] set of inaccurate or incomplete marketshare studies or trade rag articles that may get referenced in future marketshare studies. the only half-way accurate way to attempt to obtain marketshare percentages comparatively is to look at revenue, or the number of actual customers by type, and those numbers aren't typically the type that network engineers care to analyze. i suggest that you ask the DOJ, the FCC, or the EC, as I suspect that they have the most accurate response to the question as has ever been compiled. i am, however, quite interested in your overall results. best of luck, - jsb On Tue, 2 Jan 2001, Eric Lemond wrote:

The Internet Metrics post in December was a good starting point for my research on backbone traffic, but I'm struggling to get more specific information. Specifically, I am trying to find out the market share for the major backbone providers (so far I've found WorldCom 37% and Sprint 16% of traffic). Boardwatch has this breakdown, but by ISP connections (WorldCom 24% & Sprint 14%). Is the percent of ISP connections a good estimate of the percent of traffic a backbone carries?

Also, for the major backbones, is there any kind of breakdown of the percent of a backbone's traffic that is for a corporate end user vs. a personal dial-up or broadband account. Specifically, does WorldCom, Sprint, C&W, ATT, Genuity, Level 3's traffic have a strong tendency to be for corporate or personal accounts?

Finally, Is there any source or estimate for the current total amount of Internet traffic carried by all networks?

Any information or sources would be greatly appreciated. Thanks, Eric ---------------- Eric Lemond Interdiciplinary Telecommunications Program, CU Boulder Intern, FastIdeas