Re: VCC cost models .... (was Re: Limits on SVCCs)

[[e-mail]]

Eric,

> My point was that the right place to make the decision to allow for a short-
> cut is most often in the network (not the host).

My point is that the right place to make this decision is at the entity
which will (ultimately) pay the bill.

>> Interesting problem, no?
>>
>
> Only as an implementation detail...

I think it's more than just an implementation detail, it's a major shift in 
the way that people (users or the owners of edge devices) have to deal with
the costs of Internet usage.

> It may be due to my somewhat faulty intuition, but I suspect that _you_
> (whoever _you_ is) pays...

This is another interesting problem, moreso than the unicast case.  I agree
that it's unreasonable to expect the senders to pay for the call.  So, if the
MC sender (either a sender or MCS) places the call to the receiver (the
_you_ in this case), and we expect the receiver (the user or an edge 
device) to pay, this looks to me like a collect call (and in
the MCS case, who pays for the calls from the senders to the MCS?).  So if a 
receiver joins an MC group then it (or the edge device which connects it to 
the ATM network) could get hit with any number of collect calls (one per 
possible sender) as the result?   If so, this is a big change in how people 
will have to deal with the costs of operating and administering the Internet.
This is also a technical issue.  That is, what mechanism do we have for
properly screening incoming collect calls at any directly connected ATM
device?  How would any directly connected device determine which collect
calls to accept and, if passing on the charges to users, how to bill them.
MC cut-through raises a lot of really interesting (and nasty?) issues.

Again, all this really only applies to the Internet when the cost of
these connections would be high.  LAN and campus networks (where presumably
the costs are negligable or very low) would appear to be in a position to
benifit more from cut-through (unicast or MC).

 --- pete

------------------
Peter Schulter					[e-mail]
Digital UNIX Networking				voice [phone]
Digital Equipment Corp				voice (DTN) 381-2920
ZK3-03/U14					FAX   [phone]
110 Spit Brook Road				FAX   (DTN) 381-2257
Nashua, NH 03062