Re: Multiple BFD sessions over one path
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Multiple BFD sessions over one path



But the point is that, the way 1hop is specified, there is *no* way to demultiplex the initial packet--thus the requirement.

The only mechanism for multiple sessions is to have multiple paths, with one session per path. A "path" ends up with the somewhat reflexive definition of "something you can demultiplex by."

As such, I believe that the language in the two drafts is proper.

--Dave

On Sep 19, 2009, at 2:07 AM, Mach Chen wrote:

Hi Dave,

Thanks for your response!

Based on your explanation, the point is whether there is a way to demultiplex the initial patcket, if there is, means no need to limit the BFD sesion. So, IMHO, it's better to clarify this in 1hop(since the draft is still in IESG review state, seems not too late to do it:-).

Have a good weekend!

Best regards,
Mach


--------------------------------------------------
From: "Dave Katz" <dkatz at juniper.net>
Sent: Saturday, September 19, 2009 2:39 PM
To: "Mach Chen" <mach at huawei.com>
Cc: <rtg-bfd at ietf.org>
Subject: Re: Multiple BFD sessions over one path

The point of the text in 1hop is that there isn't any use in having more than one session over a "path", and in any case it can't be done because there would be no way to demultiplex the initial packet of multiple sessions on the same "path" (because the first packet carries discriminator 0).

But diffserv is essentially a tunnel of sorts; each service level has potentially different forwarding characteristics, so it may be useful to run BFD over more than one level (and the sessions can be demultiplexed because they are contextualized by the diffserv level).

Essentially, each diffserv service level is considered to be a separate path, not unlike GRE or IPsec or what have you.

Implementors are free to do as they please; the point of the text is to not preclude creative use of BFD where it can be useful.

--Dave



Note: Messages sent to this list are the opinions of the senders and do not imply endorsement by the IETF.