 |
 |
 |
 |
|
 |
|
 |
|
 |
|
 |
|
|
|
|
|
|
|
|
|
|
|
|
Hi Alper, Alper Yegin wrote:
Also, see the below text from RFC 3775: Normally, a Binding Update specifies the desired care-of address in the Source Address field of the IPv6 header. However, this is not possible in some cases, such as when the mobile node wishes to indicate a care-of address which it cannot use as a topologically correct source address (Section 6.1.7 and Section 11.7.2) or when the used security mechanism does not protect the IPv6 header (Section 11.7.1). If the reason is topological, how do you expect the MN to receive the BA sent to such topologically-incorrect IP address if HA sends the response back to the Alt-CoA, as opposed to the source address of the BU?If you continue along with this reasoning, then the MN won't receive any data packets that the HA tunnels to the topologically incorrect IP address either. So the HA can't use the topologically incorrect IP address for both the Binding Ack messages and for regular data traffic. So what was the point in the MN sending this address to the HA in the AltCoA option?The path followed by BU/BA does not have to be same as the path followed by the HA-MN tunnel. That separation is possible with the use of Alt-CoA where Alt-CoA dictates only the tunnel end-point on the MN side (not the destination IP address of the BA).
But if the MN is binding an address that it can't receive regular traffic at, then something is broken right? The BU/BAck exchange might succeed, but the MN is not able to receive any traffic that is sent for its home address. The Home Agent would be tunneling to the care-of address.
Lets step back a bit. The discussion started off with us trying to figure out where the binding revocation should be sent. RFC 3775 does not require the home agent to store the source address on the binding update if the source address does not match the address in the AltCoA option. There is only one address - the care-of address, stored in the binding cache entry. And the care-of address comes from the AltCoA option, if it is present. Data traffic and subsequent signaling messages initiated by the Home Agent would be sent to this care-of address.
The source address on the original BU might not even be valid anymore.Then we started talking about the Home Agent/LMA being able to store two separate addresses, one for the control plane and one for the data plane, I think defining a new mobility option to carry the data plane address is the cleanest solution.
Vijay _______________________________________________ MEXT mailing list MEXT at ietf.org https://www.ietf.org/mailman/listinfo/mext