Re: [Idr] BGP flooding and AS-PATH

["Ketan Talaulikar (ketant)" <[e-mail]>]

Agreed, Robert. Will capture this in the next update and thanks for your review and for pointing this out.

Thanks,
Ketan

From: Robert Raszuk <[e-mail]>
Sent: 11 March 2019 17:08
To: Ketan Talaulikar (ketant) <[e-mail]>
Cc: idr@ietf. org <[e-mail]>
Subject: Re: BGP flooding and AS-PATH

Hi Ketan,

I was not talking about TORs .. they will get all info just fine.

I was talking about same stage peers as they can have same AS number and with eBGP there may be a bit of challenge to get BGP updates accepted orignated by same level device.

And I was only waiting for you to mention allowas-in or as-overwrite knobs ;)

Can you reflect those options in the draft? Or at least in any other way address this scenario ?

Kind regards,
R.

On Mon, Mar 11, 2019, 11:41 Ketan Talaulikar (ketant) <[e-mail]<mailto:[e-mail]>> wrote:
Hi Robert,

Please check inline below.

From: Robert Raszuk <[e-mail]<mailto:[e-mail]>>
Sent: 11 March 2019 15:05
To: Ketan Talaulikar (ketant) <[e-mail]<mailto:[e-mail]>>
Cc: idr@ietf. org <[e-mail]<mailto:[e-mail]>>
Subject: BGP flooding and AS-PATH

Dear authors of draft-ketant-idr-bgp-ls-bgp-only-fabric,

I would like to ask for a small clarification on how in the quoted RFC7938 design are you going to propagate topology information between peers configured with the same AS number ? Say between devices belonging to tier 1 stage or to a sets in tier 2 stages ?

Do you assume that those switches just do not need to be aware of each other's presence say for TE purposes ?
[KT] This isn’t necessary in the use-cases where controller is doing the TE computations. E.g. a TOR can delegate end to end TE computation (from itself all the way over the DCI/WAN to a remote node) to the controller.

Specifically I am curious how are you going to accomplish it in the cases where no AS-PATH removing controller is involved as described here:


"An internal software component on any of the BGP routers

(e.g. TE module) can also receive the entire BGP network

topology information from its local BGP process."

[KT] This is a different variant of the TE use-case where the TOR itself can do TE computation. In this case, you are correct about the topology flooding via the controller when using the ASN scheme in https://tools.ietf.org/html/rfc7938#section-5.2.1 . For this use-case, there are other options and designs though (incl private AS for unique assignment to TORs, allowAS-in, etc.) to make this possible.



Thanks,

Ketan

Thx a lot,
R.