Hi Joseph, >> Used link-layer does have an impact: on cellular networks >the link-layer is nowadays most often opened on-demand (i.e. >when an application starts), or quickly reopened e.g. if >network connection is temporarily lost for any reason. In the >future always-on connectivity becomes more commonplace than it >is today, but temporal connection failures may remain in >game.. So, when a connection is (re-)opened, IPv6 address >needs to be obtained as quickly as possibly for the reasons of >good user experience (user is waiting for something to >happen..). Often the connection is trusted point-to-point >connection, so there is no risk of spoofed RAs. As user is >waiting, there is no time to start guessing whether DHCP is >available or not, but address will be configured by fastest >means available - currently by sending RS immediately when >link layer gets up to ensure fastest possible retrieval of RA. >> > >If delay time for address configuration is crucial for user, >how about DNS or other configuration informations? Are those >informations provisioned through stateless DHCPv6? Now, I >guess that stateless >DHCPv6 is initiated right after the first RA is received with >O flag set to 1. Is this delay acceptable for your environment?
Some configuration information can be received during link layer activation before IP address configuration. 3GPP 24.008 http://www.3gpp.org/ftp/Specs/html-info/24008.htm lists these, including the DNS server addresses. If a host does not wish to/is not able use these link layer means, it can use Stateless DHCPv6 for the configuration. Unfortunately RFC5006 is at least not yet mentioned in the specs. The 3GPP originally specified the link layer means I guess at least partly because at that time RFC3736 and RFC5006 did not exist. Another alternative to get DNS server addresses is statically when a host is provisioned with network connection settings. Of course RTT for a single _successful_ message exchange in modern wireless networks is rather small (especially when compared to roughly 200ms in older ones where IPv6 was first introduced), but as the number of messages exchanged by various protocols of different layers increases, the time builds up. It's best to optimize wherever possible. Best regards, Teemu -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
