Hi Fred, >> As promised, let me enumerate the "protocol" worked involved for solutions >> aimed at "discovery, >> selection, and coordinated execution of mobility protocols at multiple >> layers". >> >> - Discovering network's mobility capabilities: >> >> Whether it supports PMIP, LISP, etc. >> Possible approach is to define new DHCP options to deliver this "network >> info" to the terminal. > > Another approach is DNS. For example, nodes can discover whether the > network supports AERO by resolving the FQDN "linkupnetworks.domainname". >
Yes, that's possible too. >> - Discovering corresponding node's mobility capabilities: >> >> Whether it supports MPTCP, MIP route optimization, etc. >> Possible approach is to use DNS-based discovery. > > With AERO, it might be easier to just test the assumption that > the correspondent participates in the service and then make > other arrangements if it does not. > Trial-and-error, when fails, would create additional latency getting in the way of starting the e2e communications. >> Discovering the MN's own mobility capabilities does not involve any protocol >> work. It may be based on >> platform-specific methods, API, application profiling, etc. > > OK. > >> How the terminal selects the mobility protocol(s) to apply to a given flow, >> and how it coordinates >> execution of them are materials for an "informational" document that'd also >> refer to the >> aforementioned discovery elements. >> >> Like Danny was suggesting, we can tackle this in the working team that deals >> with the source address >> selection, as there's an interaction between the two. Whether a flow needs >> a fixed or sustained or >> nomadic IP address is influenced by whether the application traffic would >> need IP-layer mobility or >> not. > > It seems like some flows would need to go through the home network > using a stable home network address while others should go through > the visited network using an address specific to the access network. > And somehow, the mobile needs to keep the home and visited domains > separate. > Yes. > Don't some systems already do that? Or, am I missing the point? > The closest I can think of is: 3GPP systems using, say IMS APN and Internet APN, where the Internet APN uses SIPTO at the local network. Do you have this, or something else in mind? Alper > Thanks - Fred > [email protected] > >> Alper >> >> >> >> >> _______________________________________________ >> dmm mailing list >> [email protected] >> https://www.ietf.org/mailman/listinfo/dmm _______________________________________________ dmm mailing list [email protected] https://www.ietf.org/mailman/listinfo/dmm
