Document: draft-ietf-imss-ipv6-over-fibre-channel-01 Reviewer: Michael Patton Date: April 14, 2004 This is the second review cycle in a row that I've had an "IP over " to review. In both cases, I now realize, it would have been much easier to understand both documents if there had been a better high-level description, and I mean really high level as in a sentence in the intro that says " is a NBMA topology" or " is a point-to-point mesh" or something at that level. I'm not sure how to make this suggestion to future writers of IP over X documents, but that would be nice... Now, the actual review: Summary: This draft is ready for publication as a PS RFC As previously mentioned, I'd like to see a general description of what FC is. As far as I can tell, this document describes a mesh which supports port-to-port unicast and broadcast (with a comment about optional multicast). I'm a bit worried that the architecture described is N-squared requiring every node to maintain a separate connection to every other node and manage state for that connection (called Exchanges in the FC lingo). This is a general problem of mesh topologies, so I'm not sure there's much that can be done, but I'd like to see a mention of this aspect. Hopefully, FC fabrics will be able to handle that quantity of Exchanges. Furthermore, it explicitly disclaims as "outside the scope of this document" the techniques for dealing with that. In section 3, it might be nice if they explained why only those specific N_Port_Name formats. This restriction is mentioned elsewhere as well. Given how much explanation of the underlying technology they have in Section 2, this oversight seems odd. After all, they laid the groundwork for all the other things on the list. I did see some explanation in 6.1, with details in the rest of section 6, so the simple fix is just a forward reference from Section 3 to Section 6. Although mentioning the restriction and why it's acceptable a little earlier than section 3 might be nice. The acronym ESP applies both at the FC layer and the IPv6 layer. What it means at the FC layer is not explained (nor is its meaning at the IPv6 layer, but I know that one). From context it seems reasonable to assume that they are, in fact, similar, but that should be mentioned. Now, given my assumption that they are similar, I see them both mentioned in section 4.1, mentioning that either may be used for security. Yet, there is no discussion, either in that section or in the Security Considerations section, of what the tradeoffs are for choosing one or the other. [I actually know, but think it should be explicitly stated in the document somewhere.] The last paragraph of Section 5 talks about "mixed media" but everywhere else the document seems to imply that FC is one media. So, I don't see what the heck this paragraph is about. Either it's spurious and should be removed, or better explanation of what they're talking about is needed. simple typos ------------ First paragraph of intro has a comma that makes the grouping wrong. The comma at least should be removed, but even better would be to reword the single sentence paragraph as several sentences. 4.1 "resulting to the" => "resulting in the"