Re: [bess] Adam Roach's No Objection on draft-ietf-bess-evpn-etree-13: (with COMMENT)

["Ali Sajassi (sajassi)" <[e-mail]>]

Hi Adam,

Please refer to my replies inline.

On 9/13/17, 6:19 PM, "Adam Roach" <[e-mail]> wrote:

>Adam Roach has entered the following ballot position for
>draft-ietf-bess-evpn-etree-13: No Objection
>
>When responding, please keep the subject line intact and reply to all
>email addresses included in the To and CC lines. (Feel free to cut this
>introductory paragraph, however.)
>
>
>Please refer to https://www.ietf.org/iesg/statement/discuss-criteria.html
>for more information about IESG DISCUSS and COMMENT positions.
>
>
>The document, along with other ballot positions, can be found here:
>https://datatracker.ietf.org/doc/draft-ietf-bess-evpn-etree/
>
>
>
>----------------------------------------------------------------------
>COMMENT:
>----------------------------------------------------------------------
>
>Section 3.3.2 says:
>
>   The PEs implementing an E-Tree service need not perform MAC learning
>   when the traffic flows between Root and Leaf sites are mainly
>   multicast or broadcast.
>
>Does this mean to say "mainly"? I would have expected "only", as in
>section
>4.3.  In particular, if "mainly" is correct, I'm unsure how unicast
>traffic is
>supposed to be handled. Is it simply flooded out (modulo filters) in the
>same
>way as broadcast traffic? If that's the intention, I think some
>additional text
>here saying as much would be useful.

Ali> Added the following sentence:
Ali> "In such scenarios, the small amount of unicast traffic (if any) is
sent as part of BUM traffic."
>
>----
>
>Section 5.1:
>
>   The reserved bits SHOULD be set to zero by the transmitter and SHOULD
>   be ignored by the receiver.
>
>The "SHOULD" here seems that it might make assigning meaning to these
>bits in
>the future problematic. If implementations decide to either assign local
>meaning to these bits, or decide that they don't need to be initialized,
>then
>future IETF specs that try to use them might be in for some pretty nasty
>deployment surprises. If these need to be "SHOULD" instead of "MUST,"
>please
>add some motivating text to the document for the sake of people who might
>want
>to extend the protocol in the future.

Ali> changed the second ³SHOULD² to ³MUST²:
Ali > "The reserved bits SHOULD be set to zero by the transmitter and MUST
be ignored by the receiver."
>
>----
>
>The IANA handling of "Composite Tunnel" seems problematic: although
>several
>values in this "Reserved for Composite Tunnel" range have well-defined
>values
>(e.g., 0x81 means "RSVP-TE P2MP LSP with composite tunnel"), they look
>unallocated/reserved in the resulting table. I think what you really want
>to do
>here is update the introductory text for the table to make it clear that
>values
>now take the range 0x00 - 0x7F and modify 0x7B through 0x7F as you've
>proposed
>doing.

Ali> updated the section to make it more clear - please refer to rev14 of
this draft. The only changes for this document is for the range of
0x7B-0xFA which was previously unassigned. The decomposition of this range
is explained in the IANA section.
>
>On top of this, I have the same concerns as Warren does regarding the
>impact of
>this change on in-the-field use of experimental tunnel types. I think the
>only
>reasonable way to retrofit this mechanism onto the existing system would
>be to
>to say that the "Composite Tunnel" bit MUST be ignored for tunnel types
>0x7B-0x7E, and possibly allocate some additional experimental codepoints
>(e.g.,
>0x77-0x7A) so that people can run experiments with tunnel types that also
>include composite tunnel behavior.

Ali> There shouldn¹t be any impact. The current tunnel types are in the
range of 0x00-0x07 [RFC7385]. The max range for the future will be in the
range of 0x00-0x7A. The mirror image of this range with the composite
tunnel type would be in the range of 0x80-FA. There is complete backward
compatibility with existing experimental values.
>
>