> > => I took out the RFC text to make the mail shorter. 
  > > When I made that comment I meant to distinguish IPv6
  > > nodes from MNs based on whether they implement MIPv6
  > > MN functions or not. So when I said "stationary" I meant
  > > IPv6 nodes that do not implement MN functions as described
  > > in MIPv6.
  > 
  > OK, though choosing to call such nodes "stationary" isn't very
  > intuitive. My "stationary" laptop goes with me to lots of 
  > places, yet
  > would be a good example of where temporary addresses might well be
  > appropriate.

=> ok, my point was not related to physical mobility, but 
topological mobility. Of course, you're saying that topological 
mobility does not imply session continuity and consequently
does not imply MIPv6. 

  > > For those MNs, we do not yet have a mechanisms
  > > to allow for RFC3041 type home addresses. So there would be
  > > no point in having them for the CoA since the HoA is always
  > > visible for traffic analysis.
  > 
  > Specifically, I assume you mean that there is no way for an 
  > MN to use
  > temporary addresses because there is no way for the MN to 
  > tell the HA
  > what other addresses it is using?

=> Correct. There is no IETF _specified_ way for this yet. Of course
one can think of ways of solving this problem for some scenarios 
but there is nothing specified today. 

  > 
  > Also, how much of a problem is this perceived to be?

=> Which problem? The fact that the mobile node cannot 
generate 3041 home addresses? 
If so, it's a significant challenge. I can think of 
several ways of addressing this problem but they are not 
free and I don't think they solve this problem for 
all possible scenarios. 
For example (not an exhaustive analysis at all): 

- CGAs can fix this problem. However, frequently generating
PKs is computationally intensive. Also, HAs are likely to
need a higher level identifier for the MN in order to offer
their forwarding services. For instance, an ISP offering a HA
"service" to its MNs is likely to not want every other MN
to use its HA, when it hasn't paid for this service. 

- If a HA is configured with every possible "allocated"
HoA. It might be able to accept 3041 HoAs. By "allocated"
I mean that it is used to allow MNs to be reachable. But 
this idea will certainly need significant specification
work and cannot be generalised for all scenarios. For example, 
how does a HA know if a MN has updated the DNS with its 
new 3041 HoA? What happens when the MN stops using 
that address? What happens if the MN goes out of coverage
for a period where the binding times out and another one
takes that address? 

Perhaps with enough restrictions and changes to existing 
specs these problems could be solved, but they're not trivial. 

Hesham




--------------------------------------------------------------------
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]
--------------------------------------------------------------------

Reply via email to