[PATCH] BGP: Add support for BGP hostname capability

[Job Snijders]

Dear Vincent,

Thank you for your contribution, running code always is an excellent way
to move specifications for the purpose of interopability forward.

I'd like to comment on my favorite topic... *** DEFAULTS **** :-)

On Wed Feb 3 19:19:50 CET 202, you wrote:
>
> However, maybe the capability should not be enabled by default.
>

Indeed, it should absolutely not be enabled by default. Enabling a
feature like this would violate the principle of least astonishment! :)

Other than POLA, there are few more reasons it probably shouldn't be
enabled by default: i recall the draft's supporters alleged this feature
is useful in context of 'datacenter networks' (aka, 'non-Internet'?).
This to me means that as only a small subset of users care about it, and
the onus is on them to enable it if they wish to use this feature.

The draft being in draft status, and not (yet?) been adopted by the IETF
IDR working group is another reason not to enable it by default.

Thirdly, the semantics of the mechanism are not entirely well aligned
with the mechanics of a node's hostname. A node's hostname can change
(often?) over the lifetime of the node, while signaling the hostname
through a BGP capability effectively limits hostname exchange merely to
the OPEN event, a single point in time.

Lastly, many have argued that this is what we have Reverse DNS for, and
then others argued that in ISIS the hostname also is flooded, so why not
in BGP... and so on, making it clear there is no consensus what to do
next.

I recommend adjusting the patch in such a way that the capability is
only exchanged with specific neighbors where the capability has been
explicitly enabled through neighbor/group specific configuration.

Kind regards,

Job