Re: Mobile IP and Mobike

 In your previous mail you wrote:

   >From the discussion on the list and at the meeting this week, it sounds like
   Mobike and Mobile IP are separate and incompatible solutions to mobility,

=> they are incompatible as competing signaling between the mobile
node and the home agent, using the mobile IPv6 terms. One can use
the routing optimization of mobile IPv6 with a correspondent node
when mobike is used for the signaling with the home agent.
So the context (again) does matter.

   with orthogonal deployment scenerios. They are incompatible because
   deploying them together on the same box would result in duplicate "binding
   update" messages between the host and the tunnel endpoint (which is the home
   agent in Mobile IP).

=> I can't disagree (:-).

   From this, I deduce that having both in a host stack at
   the same time is to be discouraged.

=> I disagree: even the home agent and the security gateway can have
the same role they are deeply different. In fact, IMHO this convergence
between security and mobility is a very good thing, at least it should
obsolete my old I-D asking to avoid double tunelling!

   Their deployment scenerios are orthogonal because Mobile IP is
   targetted at wireless network service providers

=> this mobile == wireless == cellular credo makes no sense.

   and Mobike is targeted more at enterprise customers with road warriors.

=> in this room (SAAG session at IETF meeting) many use a nomadic VPN
(aka road warrior VPN) over a wireless network. I don't believe
there is a real split in usage, just that users will come from
two different communities.

   There might be some overlap for cases such as UMA, which is using
   IPsec tunnels set up using IKEv2 over WLAN to tunnel gateways through which
   circuit-switched GSM voice traffic and GSM signaling is directed, in
   addition to GPRS traffic.

=> I can speak about GPRS: the IP service is so poor (private IPv4
addresses, filtering madness, etc) that the best way to use them for
real things (i.e., to go somewhere throught the Internet) is to use
L2TP, i.e., use one of the options offered by current VPNs.
But mobility is not forgotten because it is needed for "vertical
handovers" when you can get locally faster and cheaper than GPRS.

   Am I missing something? Is more detailed analysis necessary to confirm this
   admittedly superficial description? Fire away!
   
=> orthogonal means than the product is zero. IMHO this is far too
pessimistic, and for the MOBIKE side you forget the multi-homing
and SCTP contexts (do we need to change the name of the WG to
avoid this bias?)

Thanks

[email protected]