At 12:52 PM 10-10-96 -0800, you wrote:

| would you mind expanding a bit on why SVCs would do the job? | We're trying to determine if it's a service worth offering anytime | soon, who would use it, etc.,

Basically SVCs are easier to support and more robust than PVCs, even if all you do is nail them up like PVCs. As an example, say an ATM service provider sets up a L2 service and a bunch of ISPs come and begin to interconnect bi-laterally. Each PVC has to be coordinated among three parties -- two ISPs and the ATM service provider. This is time consuming and error prone, but workable. Now, suppose after all this has been set up, something happens to the ATM switch and it forgets about a PVC. (Of course this would never happen, just like no router would ever get in a strange state and forget a route.) There is nothing on the endpoints able to re-establish the circuit, even though it is obvious that it is down or missing. No one on the endpoints can change the QoS or reconfigure their circuits and no one can set up their connection to accept new connection requests on demand, etc. In short, even without thinking about futuristic bandwidth on demand, SVCs would make the current classical paradigm easier to manage and more robust. Your first customer is your current service delivery department and NOC. --Kent