[e2e] Re: crippled Internet
braden at ISI.EDU
Wed Apr 18 14:51:02 PDT 2001
*> responses to DoS attacks get sent? Who ends up paying for them?),
Actually, the last is easy: the administrators of the nodes that were
"stolen" to mount the DoS attack pay for the traffic generated by
their nodes. This provides the right incentive structure, IMHO.
*> legislation to prevent that fraud (since you're no longer using up
What fraud? If you send a packet, you pay. If you don't pay, your
certificate of access is revoked. (Large amount of hand-waving here,
*> someone's leased capacity, but actually stealing from their provider/
*> phone company), and resulting stronger regulation of services -- which
*> discourages new applications.
*> And then value-added charging gets implemented, where UDP packets cost
*> far more than TCP packets, since anyone using UDP is obviously up to
*> no good. ICMP packets will be _really_ expensive.
Why should the network care about UDP/TCP/SCTP/whatever transport? It
would charge for IP datagrams.
*> If the world's going to move to usage-based charging, it's going to
*> have to implement some kind of widespread authentication model
The hard question you didn't mention is how to divide the cost
between the two ends of the connection.
More information about the end2end-interest