On Tue, 31 May 2011 11:29:49 +0200 Markus Hanauska <[email protected]> wrote:
> > On 2011-05-30, at 22:05 , Ray Hunter wrote: > > > Which source address (SLAAC/DHCPv6) would be used by the client for an > > outbound session if a SLAAC address and a DHCPv6 were both configured on > > the same link and with the same prefix, in the absence of a flag? > > As I already said in my previous mail: The DHCP address should win by > default; that is, of course, unless the admin/user of the node has configured > anything else. If a network runs a DHCP server that hands out addresses, > network admins probably do so for a reason. The SLAAC one is valid and > inbound traffic might arrive for it, but for outgoing traffic, some RFC > should simply have given an order of preference: > > 1. Manual configured IP > 2. DHCP > 3. SLAAC with Privacy Extension > 4. SLAAC with Interface ID > Some people might prefer SLAAC over DHCP. For example, if I were attached to a public hot spot with my laptop, then I'd want to use SLAAC with Privacy Extensions in preference to DHCP, even if there is a DHCP server available. > > Think dynamic DNS too: which (destination) address(es) would be auto > > registered at the server end? > > All of them, since a single DNS entry can resolve to plenty of different IP > addresses (DNS has always been designed to allow this scenario). The order > should be the same as the order of preference above, though you can argue if > it is meaningful to register Privacy Extension addresses at all, since they > will dynamically change pretty often. > > > There are a lot of operational people who currently rely on having > > predictable IP addresses for accounting, audit, scripting, firewall rules, > > neighbor filtering, fault tracing, reverse DNS, policy based routing, > > setting DSCP in QoS, any other number of ACL's .... > > I can also totally agree with your statement above - but there is one thing > you are missing: It is not only important, to have predictable IP addresses > in some scenarios, it is also important to have "constant" IP addresses, > since even if the address is predictable, it will be a problem if it is going > to change every couple of months. SLAAC addresses based on interface IP are > in fact predictable; but they not constant! On the other hand, when using > DHCP, it takes me 2 minutes to change the stored interface ID of a DHCP > client in the server config to assure, that even after an interface ID > change, the client still gets the same IP address as before. How will you do > this with SLAAC? Manually configuring the same MAC address as the old > interface? You shouldn't manually assign MAC addresses, if not really > necessary, especially not those which claim to be globally unique, even > though they are in fact not. MAC address collisions are worst case scenarios > for most networks. Further c > ertain clients, e.g. take an iPhone or iPad, don't allow you to change > interface IDs at all AFAIK. > > Regards, > Markus > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > [email protected] > Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
