On 9/29/2017 9:13 AM, The IESG wrote: > A new IETF WG has been proposed in the Routing Area. The IESG has not made > any determination yet. The following draft charter was submitted, and is > provided for informational purposes only. Please send your comments to the > IESG mailing list (iesg@xxxxxxxx) by 2017-10-09. ... > > Network solutions based on the concept of Identifier-Locator separation are > increasingly considered to support mobility, overlay networking for > virtualization and multi-homing across heterogeneous access networks. The problem there is that the same properties that facilitate routing also facilitate tracking. Consider a mobile node that switches from a Wi-Fi network to a cellular network. In the current state of the art, there is no relation between the Wi-Fi address and the cellular address. Intermediaries cannot observe the traffic and deduce that two different flows of IP packets originate from the same node. In contrast, with an ID/Loc architecture, the two flows are associated with the same identifier, which can then be used to track the movements of the device. Similarly, consider a node that connects several times to the same network, and each time uses IPv6 temporary addresses. The web servers that it contact cannot use the IP addresses to correlate different connections that happened at different times. This would change if the identifier in an ID/LOC architecture remained constant. Multipath TCP and planned multipath extensions of QUIC are example of transport protocol that allow transport connections to use multiple network paths simultaneously. In both cases, there s significant work going on to ensure that intermediaries cannot easily associate the traffic on the multiple paths with a single connection. If the multi-homing function was delegated to an ID/LOC system, intermediaries could potentially observe the identifiers and associate these connections. In short, careless applications of the ID/LOC architecture could easily result in serious privacy issues. The proposed charter does include a brief statement about privacy: > - Analysis of the concepts of identity-identifier split and dynamic > identifier changes, including their implications on anonymity and privacy. > Explicitly, the framework must define privacy requirements and how potential > extensions/solutions should meet them. This is a good start, but the whole concept of "unique identifiers" is scary, and I would like to see this expanded. For example, I would like to see an explicit reference to a baseline, e.g. assuring no privacy downgrade compared to IPv6 temporary addresses, or assuring that hosts that elect to not be tracked when roaming across networks will not be. I also know that there have been discussions of hiding identifiers from intermediaries, and i would like to see that as an explicit goal of the proposed WG. -- Christian Huitema