Fred Templin <[EMAIL PROTECTED]> wrote: |Brian, | |For the most part, agree with all of your points of clarification; |thanks. One item for further discussion: | |Brian Haberman wrote: | |>> |>> Long period of time, perhaps, but how long is difficult to |>> quantify. Again, perhaps this is a question for the [BAKER] |>> renumbering draft. |> |> |> I think it is quite pertinent to this draft. How long must an |> address remain stable in order to satisfy this document's stability |> goal? | | |I think this question is going to be very much dependent on |the use case scenarios. For example, in the future we may see |deeply-networked homes in which we will have long-lived |sessions, such as playing a 3hr video between the DVD player |and home theater system. Indeed, there may be "always-on" |sessions such as the home-security system monitoring |intrusion-detection sensors around the house.
I continue to believe that there is more to the notion of stability than can be expressed by a simple time scalar. Any measure of stability must also take into account the ability of anyone other than the user/owner of the address to force a change either directly or indirectly. As soon as you start to consider specific time frames as the primary characteristic of stability you almost implicitly admit to a model where the control of the address space (probably even in the short term) rests with a third party. This in turn subjects the user of the address space to the technical problems of the controlling entity and (perhaps worse) to the ability of that entity to charge arbitrarily for continued stability. There may be a second-order effect here, i.e., how long is a particular stability time frame valid? Today, my ISP might be willing to guarantee me a minimum of 3 hours stability for $x. But if next month it can change the price for 3 hours of stability to 100*$x, was the address space ever really stable even by the measure of a 3 hour time frame? Ultimately I think stability is much closer to a binary property than to something that can be measured in hours. It may not (or may) be possible to guarantee local address stability in the face of global change (e.g., to a new version of IP), but to tie stability to almost anything of a smaller scope is in effect to remove stability. In particular, anything that depends on the promise and competence of the user's service provider (which in turn depends on the promises and competence of all the service providers up the line) cannot realistically provide stable addresses. There doesn't seem to be any obvious reason other than third-party control to consider specific stability time frames. Therefore, I suggest that the only reasonable life for a stable address is, to a first approximation, infinite. Dan Lanciani [EMAIL PROTECTED] -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------