The BGP peering issues that have recently come up really come directly from a disagreement about what constitutes network value. I think by now we all realize that peering is a sort of quid pro quo, where both parties get some value from the deal. Traditionally, this was an easy thing to assess. Large backbone providers tend to have a a nice mix of dial-up users, content providers, and everything in between. Measuring the "value" they bring to the table could easily be qualified by a traffic study. In the brave new world this isn't the case. One can easily argue that a dial-up only provider and a content hoster should peer. Each brings value to the relationship, but the traffic distribution will be highly asymmetrical. In this situation what metrics do you use to assess the value of the peering arrangement? Is there a good metric like number of users, or number of flows that you can use to measure value? Does value have to be something that only a human can determine with lots of subjective measures? Has anyone tried to qualify the value of a network for such purposes? I'd be interested to know the details about that process. If you are able to find a value, can you then move to a "price / performance" ratio by determining the cost of the peering session? I suspect this is more difficult to determine than the wealth of the network, at least over shared peering points where the direct cost for a single peer is not a line item. Possibly the most difficult question of all, is there any value to everyone using the same metrics? Ignore the problems with creating a suitable metric, but if one existed that ranked all providers from say, 1-10, do you think it would have a positive or negative effect on peering arrangements? I welcome public and private discussion, but hope to avoid any current issues (for fear of a flame war). :-) -- Leo Bicknell - bicknell@dimension.net Network Engineer (CCIE #3440) - Dimension Enterprises 1-703-709-7500, fax, 1-703-709-7699
On Wed, 12 Aug 1998, Leo Bicknell wrote:
distribution will be highly asymmetrical. In this situation what metrics do you use to assess the value of the peering arrangement?
Value involves money so that is the best metric. Do the backbone customers pay money for access? Do the web hosting customers pay money for access? Yes in both cases so the problem is reduced to engineering the best network to supply that quality of access that the customers have paid for. Maybe GTEI sees itself as a bargain basement cheap low quality provider and wants to dominate that market niche? -- Michael Dillon - Internet & ISP Consulting Memra Communications Inc. - E-mail: michael@memra.com Check the website for my Internet World articles - http://www.memra.com
The "value" of traffic exchanged is going to take a back seat to clout. In other words, when a $30 billion company with 20 million customers (for example) is negotiating with a $250 million company with 5000 customers one of them is going to be able to dictate terms to the other, and the actual value of the traffic measured in a vacuum (i.e. NANOG) will most likely be irrelevant. Contrary to the altruistic opinion of many, the end game for this industry subject to unfettered free market capitalism isn't pretty. The natural tendency of large cap companies in the absence of regulation is merger. As those companies consolidate, you can expect the backroom negotiation with other companies to become very heavy handed. Ignoring market capitalization for a moment, and instead assume Internet clout is measured by number of leased line customers, dialup ports, or traffic volume. BBN is initiating/facilitating a trend that most likely has them at a disadvantage since they are a minority market share holder by any of those measurements. One can only asume that GTE figures they will be much bigger in future and will gain their rightful position in the heirarchy. heh. Value has nothing to do with it and is almost pointless to talk about, except perhaps to network operators concerned about delivering the best service to their customers. Mike. On Wed, 12 Aug 1998, Leo Bicknell wrote:
The BGP peering issues that have recently come up really come directly from a disagreement about what constitutes network value. I think by now we all realize that peering is a sort of quid pro quo, where both parties get some value from the deal. Traditionally, this was an easy thing to assess. Large backbone providers tend to have a a nice mix of dial-up users, content providers, and everything in between. Measuring the "value" they bring to the table could easily be qualified by a traffic study.
In the brave new world this isn't the case. One can easily argue that a dial-up only provider and a content hoster should peer. Each brings value to the relationship, but the traffic distribution will be highly asymmetrical. In this situation what metrics do you use to assess the value of the peering arrangement? Is there a good metric like number of users, or number of flows that you can use to measure value? Does value have to be something that only a human can determine with lots of subjective measures?
Has anyone tried to qualify the value of a network for such purposes? I'd be interested to know the details about that process. If you are able to find a value, can you then move to a "price / performance" ratio by determining the cost of the peering session? I suspect this is more difficult to determine than the wealth of the network, at least over shared peering points where the direct cost for a single peer is not a line item.
Possibly the most difficult question of all, is there any value to everyone using the same metrics? Ignore the problems with creating a suitable metric, but if one existed that ranked all providers from say, 1-10, do you think it would have a positive or negative effect on peering arrangements?
I welcome public and private discussion, but hope to avoid any current issues (for fear of a flame war). :-)
-- Leo Bicknell - bicknell@dimension.net Network Engineer (CCIE #3440) - Dimension Enterprises 1-703-709-7500, fax, 1-703-709-7699
+------------------- H U R R I C A N E - E L E C T R I C -------------------+ | Mike Leber Direct Internet Connections Voice 408 282 1540 | | Hurricane Electric Web Hosting & Co-location Fax 408 971 3340 | | mleber@he.net http://www.he.net | +---------------------------------------------------------------------------+
participants (3)
-
Leo Bicknell
-
Michael Dillon
-
Mike Leber