I am an assigned INT directorate reviewer for draft-zhu-intarea-gma. These comments were written primarily for the benefit of the Internet Area Directors. Document editors and shepherd(s) should treat these comments just like they would treat comments from any other IETF contributors and resolve them along with any other Last Call comments that have been received. For more details on the INT Directorate, see https://datatracker.ietf.org/group/intdir/about/ <https://datatracker.ietf.org/group/intdir/about/>.

This document does propose several mechanisms that can increase the MTU efficiency of encapsulation protocols in the style of GRE, which seems useful. However, the document needs to be improved for clarity and safety before publication even in the Independent stream, in my opinion.

- Some claims in the abstract and introduction need explanation. A device connected to multiple networks doesn’t directly require solutions like GRE that use additional encapsulation; many devices connect without this. Instead, this solution really is isolated to overlays across networks. This needs to be clarified for scope early on.

- The GRE references are to an Independent submission of Huawei’s version of GRE. It seems misleading to not be referencing RFC 2784 or RFC 2890 directly.

- Section 4 should be broken into multiple sections, one for each format; it is difficult to understand where the details for each mode overlap or contrast.

- Many of the reference to IP headers seem to assume IPv4 (such as the IP checksum, not present in IPv6). Any document coming out now should be designed with IPv6 in mind first, and I would suggest breaking out the examples for both IPv6 and IPv4 separately. Similarly, any UDP encapsulation mode needs to be given as a complete example, not just an aside.

- Section 5, on fragmentation, may run into some of the problems with fragments in general. Please see RFC 8900. The recommendation is to either remove the fragmentation support, or strongly discourage it and reference RFC 8900.

- Similarly, concatenation as described in Section 6 may be better handled at higher layers. QUIC, for example, allows packing multiple packets in single UDP datagrams.