Re: [Idr] WG adoption call for draft-abraitis-bgp-version-capability-08, to end September 25

> these times BGP is used not in telcos mainly.

Great ! Thanks for this valuable insight.

Well last time I checked BGP is rather needed for the Internet. And this is
major focus here to keep The Internet running.

If we start throwing one off stones into it - it can still hold ... till
the time when it no longer can. And then what ? Do we have backoff plan ?
What if someone takes 128 bytes of capabilities tomorrow ?

I proposed splitting BGP into telco/Internet BGP and all other use cases
many year ago. All you enumerated fit perfectly into the latter.

- - -

Donatas,

Said all of the above - I do like your suggestion to augment BGP with this
extra information. It may be useful. I would actually still like to know
who is behind an RR as far as BGP implementation is concerned though.

What if we define new message (we can call it operational message) and for
now just have one TLV there carrying OS version of BGP between connected
peers ?

What do you think ? Would you add this to FRR ?

Many thx,
Robert.

On Wed, Sep 9, 2020 at 7:28 PM Donatas Abraitis <
donatas.abraitis@hostinger.com> wrote:

> Robert, I fully understand you, but these times BGP is used not in telcos
> mainly.
>
>  It's happening in services. Even MySQL HA can be achieved with help of
> BGP protocol.
>
> Kubernetes, Openshift, containers, etc., changed a shape how processes are
> running these days actually. But that's not about that.
>
> Imagine a cluster of 1000 containers each running a specific version of
> BGP daemon. With such setup I would need another 1000 containers running
> lldpd and exposing information about the versions. Even this won't work
> because of network namespaces.
>
> Implementing huge rfcs take time, and this is very trivial to implement
> for any existing daemons. I'm not referring big vendors here, but more
> about software vendors.
>
>
> Sent from my iPhone
>
> On 2020-09-09, at 20:16, Robert Raszuk <robert@raszuk.net> wrote:
>
> 
> Jakob,
>
> I know you love simple features easy to implement. We all to pick like low
> hanging fruits.
>
> But I support Randy in his opinion expressed on this thread that we should
> look at the bigger picture here instead of troughing ASCII or UTF strings
> at random places into current BGP protocol just because this is "easy to
> do".
>
> Thx,
> R.
>
>
>
>
>
>
>
>
> On Wed, Sep 9, 2020 at 5:32 PM Jakob Heitz (jheitz) <jheitz=
> 40cisco.com@dmarc.ietf.org> wrote:
>
>> No.
>>
>> This is a simple feature to allow an operator to get a quick idea
>> with "show bgp neighbors" what he is looking at.
>> Let's not boil this too much.
>>
>> Regards,
>> Jakob.
>>
>> -----Original Message-----
>> From: Nick Hilliard <nick@foobar.org>
>> Sent: Wednesday, September 9, 2020 1:23 AM
>> To: Jakob Heitz (jheitz) <jheitz@cisco.com>
>> Cc: John Scudder <jgs@juniper.net>; IDR List <idr@ietf.org>; Donatas
>> Abraitis <donatas.abraitis@hostinger.com>
>> Subject: Re: [Idr] WG adoption call for
>> draft-abraitis-bgp-version-capability-08, to end September 25
>>
>> Jakob Heitz (jheitz) wrote on 09/09/2020 02:39:
>> > Nevertheless, even with all these possible issues, I think it would
>> > be super handy to run "show bgp neighbors" and see the software vendor
>> > and version for each neighbor.
>>
>> yeah, this would be useful.  Would there be value in splitting the host
>> stack name from the version name?
>>
>> Nick
>> _______________________________________________
>> Idr mailing list
>> Idr@ietf.org
>> https://www.ietf.org/mailman/listinfo/idr
>>
>