> -----Original Message-----
> From: Thomas Narten [mailto:[EMAIL PROTECTED]]
> Sent: Monday, September 17, 2001 11:17 AM
>
> > > The last paragraph in section 3 is about lifetimes.
> > > I don't understand what the intended effect is of the statement
> > > since I don't know what the lifetime is of a multicast address.
>
> > See RFC 2908.
>
> 2908 seems to say mcast addresses have a start and end lifetime. That
> makes sense. What I think is potentially significantly different with
> the approach described in *this* document is that the lifetimes are
> much shorter. I.e., it is conceivable that unicast addresses will be
> advertised with prefixes of several days or a week in common
> cases. That doesn't mean the addresses will expire then, just that
> short unicast lifetimes are being advertised, just in case, it becomes
> necessary to deprecate the address relatively quickly. However, if
> multicast addresses derive their lifetimes from the same lifetime,
> there may be a mismatch in expectations.
How so? One can renew multicast addresses.
(Conceivably an implementation could support the ability to
automatically renew multicast addresses as long as the unicast
address stays valid.)
If the multicast address is being advertised with say SAP,
each SAP advertisements can contain the extended lifetime,
just like RAs do for unicast prefixes.
> Seems to me that it might very well cause problems for multicast
> applications that request an address with a start and end time that
> don't match the advertised lifetimes of unicast addresses.
If so, the same problems exist with any SSM channel that an app uses.
> I wonder if this point should to be discussed in the document
> explicitely. I.e., seems like this is a potential difference between
> IPv4 multicast and IPv6 multicast. We should be making such
> differences very clear to minimize confusion, if nothing else.
I don't think there's any difference between IPv4 and IPv6 multicast,
can you elaborate? (By the way, you should see a uni-based-mcast
draft for IPv4 sometime before next IETF.)
> Will the tying of mcast adddresses to unicast lifetimes put additional
> pressure on using longer unicast lifetimes? Is this a good thing?
I don't know.
> What are typical start and end times that folks use in IPv4 today?
Anywhere from 1-hour duration, to maybe 6 months or so.
> Do they match expectations for unicast lifetimes in IPv6?
One option might be to make it a SHOULD NOT (for advertised multicast
lifetime to exceed known unicast lifetime) with the note
that after the unicast lifetime, there is no guarantee that packets
will reach their destination.
-Dave
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
