RE: Comments on draft-bonica-6man-comp-rtg-hdr-01

[Ron Bonica <[e-mail]>]

Tom,

I'm afraid you are looking at a very old version of the draft. These fields are all gone in the current version (21).

                                                                  Ron



Juniper Business Use Only

-----Original Message-----
From: ipv6 <[e-mail]> On Behalf Of Tom Herbert
Sent: Saturday, May 16, 2020 11:31 AM
To: 6man <[e-mail]>
Subject: Comments on draft-bonica-6man-comp-rtg-hdr-01

[External Email. Be cautious of content]


A few comments...

The CRH header has three reserved bytes after the Com field presumably for aligning the SIDs. Three bytes would be needed if SIDs are thirty-two bits, but for sixteen bit SIDs only one byte of padding is needed, and in the case of eight bit SIDs no padding is needed. Since the routing header needs to be eight bytes in length, the minimizing padding for alignment can save a fair number of bytes. For instance, if there is just one 16 bit SID in the list or one to three 8 bit SIDs, without minimal padding for alignment the EH is sixteen bytes in size, with minimal padding it's just eight bytes.

I suggest considering switching the Com and Segments field since Segments is a numeric quantity field and there's some reasons to put that in the lower order bits. Masking a byte to get a numeric value is easier to conceptualize than a shift (albeit probably same performance), and also if we ever decide that the Com field needs to be extended four bits that's as easy as just reducing the halving the maximum value of segments field.

I suggest the value 0x3 in the Com field be reserved for an "extended header" format. For instance, if someday we find we really need flags in the header then an extended format could be defined that includes a set of flags, SID size, and SID list. This probably doesn't need to be mentioned in the draft, but maybe the Com field should be renamed to Fmt to indicate it describes the format of the header following the field.

>From the draft "Therefore, the CRH MAY be padded with zeros." and
"Reserved - SHOULD be set to zero by the sender.". Should both of these be a MUST?

Tom

--------------------------------------------------------------------
IETF IPv6 working group mailing list
[e-mail]
Administrative Requests: https://urldefense.com/v3/__https://www.ietf.org/mailman/listinfo/ipv6__;!!NEt6yMaO-gk!WbRs1TMhGzlPNNHDie9q_XhFrEVt5srbU8_-5vh93TMWIzwrBNLCzd6bwvTFmVsW$
--------------------------------------------------------------------