Re: [dhcwg] Stephen Farrell's Discuss on draft-ietf-dhc-dynamic-shared-v4allocation-07: (with DISCUSS and COMMENT)

[Ted Lemon <[e-mail]>]

On May 28, 2015, at 8:22 AM, Stephen Farrell <[e-mail]> wrote:
> So just to be clear: the reason I'm asking about this is mainly
> that I want to be sure that this draft doesn't make deploying
> the privacy one harder/impossible.

There are two points to keep in mind about this.   First, I think it actually makes it easier, because now the client identifier is required and the MAC address is not needed, so at a minimum it's at least easier to avoid including a consistent identifier in requests that might happen on different networks.   But secondly, the target for this specification is really home gateways, and the issues of roaming from network to network are at least at present pretty minimal.   Whether home gateway vendors implement the privacy doc or not, I think the risk to end users' privacy is limited since the upstream ISP presumably already knows who the customer is.

As for the port set ID as an identifier, it's just the opposite: it's the thing being allocated. Think of it as an IPv4 address, only with more bits.   It identifies the address and port range that has been allocated, not the client to which it has been allocated. It is sent by the client as an indication of which port set was previously allocated to the client, not as an identifier.  The client can be expected to have different port sets at different times. So it simply doesn't make sense to use it the way you are suggesting.