Michael Shields wrote:
As a router manufacturer i can assure you that keeping detailed transfer records at today backbone's speeds is totally infeasible.
Why would it need to be done at the backbone?
Because edges do not have information about the topology and routing, and so cannot asses distance-sensitive charges. Distance-insensitive per-bit charging is certainly feasible and is being practiced by some ISPs (typically as "burstable" T-1 or T-3 service). In fact, even those charges aren't likely to stand against close legal scrunity. Example of a scenario of a rationale for action against ISP X - they count received and transmitted packets on a user link. Unfortunalely, they also count packets they didn't deliver due to some packet loss in their backbone. Bingo - they charge customers for service not provided. This is fraud, pure and simple. Sorry if i gave the idea to lawyers...
Isn't it possible to have a statistically valid sampling of packets even at the highest speeds?
Good luck doing billing based on statistical sampling.
You'd know better than I would. I hope the answer is yes, because that is valuable data even if not used for billing.
Yep, that is somewhat useful data. Fortunaltely, you do not need to identify particular customers for the purposes of traffic engineering, so the problem is much simpler. Unfortunately, traffic matrices are rather useless in the Internet world - because congestion control nicely compensates for overloaded paths. I've seen T-3s getting introduced into a not particularly overloaded-looking T-1 backbone, just to get filled right away. BTW, I have a (relatively) simple solution to the Internet traffic engineering problem, but you'll have to sign an NDA if you want to know it. I already have a long list of people who "forgot" to give me any credit for things i invented. --vadim