Hi Wojciech, two comments on the 3gpp sections of draft-dec-stateless-4v6-02.
First, the definition of "Rx" is missing in section 4.2.1. Second, the table in section 4.2.2 claims "No discernible impact" for "AF Application Function" for the 4V6 translation mode. This is not fully correct. The PCC architecture support a plurality of AFs communication over Rx with the PCRF, and most of them "interact" in some way with the end user payload. An AF might terminate, proxy or transparently inspect a subset of the UE traffic (for instance control signalling, like SIP). An AF might also terminate, proxy or transparently inspect all UE traffic (application server acting as AF, deep packet inspection device, ...). And the most important point: An AF might located in different locations inside the operators networks, or even outside the operators network. Referring to your own "Figure 2" in the draft, this means that for IPv4 traffic an AF might "intercept" traffic between the "PDN-Gw" and the "S46 Gateway", i.e. with an IPv6 header, BUT it might also intercept traffic behind the "S46 Gateway", i.e., with an IPv4 header. I propose to change the impact to "No impact for IPv6. Feature to map IPv4-IPv6 addresses may be needed only in case of IPv4-only applications". I'm aware that it would be theoretically possible to "hide" this impact from an AF by using a more complex PCRF implementation that "translates" IPv4 addresses in Rx message (including TFTs) into the mapped IPv6, but this would imply limitations, and would just be one implementation possibility. I believe that this draft should not assume a specific implementation, unless clearly documented. br, Paco _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
