Hi,
Masataka Ohta wrote:
Hi Greg,
Anyway, back to the original topic, how do you think IPv6 hosts should be configured with DNS server addresses?
I think you and Pascal are saying it's not ND but PPP. Right?
Actually, I think that DHCP is better (for recursive DNS configuration).
Not so bad.
Though PPP already invoves exchanges of so many packets, DHCP does not make it worse, at least not so much.
PPP was previously used for an (in my opinion) inappropriate number of non-network oriented configurations.
I really think that the PPP configuration should concentrate on what's needed to get packets traversing the L2 link.
Yup.
However, with the current DHCPv6, it means that IP address should be configured by DHCP with four messages.
I'm not so clear on your intention here, but I'd guess that Stateless Address Autoconfiguration is OK, if there is sufficient robustness in the (re)transmission of DAD NS's and NAs.
Since these procedures need to be done for DHCPv6 configured addresses, theres no need to mandate DHCPv6 for address configuration in general (although it may be a useful access-network policy).
The OtherConfigFlag from RFC2462 (learned from router advertisement) provides another way of triggering DHCPv6 for (non-address) host configuration. From my reading of RFC3315 this will use Information-Request packets.
Although someone who knows about DHCPv6 may know better.
Whether this is time critical or not depends on the requirements on the subsystems in question.
I'd guess that a host could solicit for router advertisement and get a response within a few seconds, DNS service could be configured subsequently, after another couple of seconds using DHCP, and it may not be a problem (since it is new connections which typically need this service, rather than existing ones).
In this case, it may not be on the critical path for a configuring host.
So, I think it's better to let DHCP directly use beacon capability of the link without PPP tax (on MTU, on packet exchanges and on politics to standardize PPP over WLAN (discovery mechanism of PPPoE is useless, here)).
Perhaps the overhead is higher, although the MTU needn't be a burden if there's no switched ethernet behind (or the switching and AP support jumbograms). 802.11's MTU is much greater than 1500 octets.
Since in 802.11b the phy/mac packet headers are at least 96us (sometimes 192us) the addition of a few octets due to PPP at 11mbps (since it is point-to-point) may not contribute significantly to wireless link congestion.
If people are really interested in PPP over WLAN, they'll make a case and provide experimental results...
Greg
Base station Terminal
--> 1. DHCP Advertise (broadcast over Beacon)
<-- 2. DHCP Request (unicast)
--> 3. DHCP Reply (unicast)
Masataka Ohta
-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------