[babel] About EIGRP [was: Thoughts on rfc6126bis]

Juliusz Chroboczek <jch@irif.fr> Tue, 28 November 2017 16:06 UTC

Return-Path: <jch@irif.fr>
X-Original-To: babel@ietfa.amsl.com
Delivered-To: babel@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A97991270B4 for <babel@ietfa.amsl.com>; Tue, 28 Nov 2017 08:06:21 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.901
X-Spam-Level:
X-Spam-Status: No, score=-1.901 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id owX4OX05NXVC for <babel@ietfa.amsl.com>; Tue, 28 Nov 2017 08:06:18 -0800 (PST)
Received: from korolev.univ-paris7.fr (korolev.univ-paris7.fr [IPv6:2001:660:3301:8000::1:2]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 59BDF126B72 for <babel@ietf.org>; Tue, 28 Nov 2017 08:06:18 -0800 (PST)
Received: from mailhub.math.univ-paris-diderot.fr (mailhub.math.univ-paris-diderot.fr [81.194.30.253]) by korolev.univ-paris7.fr (8.14.4/8.14.4/relay1/75695) with ESMTP id vASG6CMl020805; Tue, 28 Nov 2017 17:06:12 +0100
Received: from mailhub.math.univ-paris-diderot.fr (localhost [127.0.0.1]) by mailhub.math.univ-paris-diderot.fr (Postfix) with ESMTP id A48EDEC873; Tue, 28 Nov 2017 17:06:12 +0100 (CET)
X-Virus-Scanned: amavisd-new at math.univ-paris-diderot.fr
Received: from mailhub.math.univ-paris-diderot.fr ([127.0.0.1]) by mailhub.math.univ-paris-diderot.fr (mailhub.math.univ-paris-diderot.fr [127.0.0.1]) (amavisd-new, port 10023) with ESMTP id 6CkIYI3I-ua9; Tue, 28 Nov 2017 17:06:06 +0100 (CET)
Received: from lanthane.pps.univ-paris-diderot.fr (unknown [172.23.36.54]) (Authenticated sender: jch) by mailhub.math.univ-paris-diderot.fr (Postfix) with ESMTPSA id CF64BEB215; Tue, 28 Nov 2017 17:06:06 +0100 (CET)
Received: from localhost ([::1] helo=lanthane.irif.fr) by lanthane.pps.univ-paris-diderot.fr with esmtp (Exim 4.89) (envelope-from <jch@irif.fr>) id 1eJiOc-0004V4-IW; Tue, 28 Nov 2017 17:06:06 +0100
Date: Tue, 28 Nov 2017 17:06:06 +0100
Message-ID: <7ivahus80h.wl-jch@irif.fr>
From: Juliusz Chroboczek <jch@irif.fr>
To: Russ White <7riw77@gmail.com>
Cc: babel@ietf.org, dschinazi@apple.com, 'Donald Eastlake' <d3e3e3@gmail.com>
In-Reply-To: <04cc01d3679d$136c56f0$3a4504d0$@gmail.com>
References: <04cc01d3679d$136c56f0$3a4504d0$@gmail.com>
User-Agent: Wanderlust/2.15.9
MIME-Version: 1.0 (generated by SEMI-EPG 1.14.7 - "Harue")
Content-Type: text/plain; charset="US-ASCII"
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.2.7 (korolev.univ-paris7.fr [194.254.61.138]); Tue, 28 Nov 2017 17:06:12 +0100 (CET)
X-Miltered: at korolev with ID 5A1D8974.002 by Joe's j-chkmail (http : // j-chkmail dot ensmp dot fr)!
X-j-chkmail-Enveloppe: 5A1D8974.002 from mailhub.math.univ-paris-diderot.fr/mailhub.math.univ-paris-diderot.fr/null/mailhub.math.univ-paris-diderot.fr/<jch@irif.fr>
X-j-chkmail-Score: MSGID : 5A1D8974.002 on korolev.univ-paris7.fr : j-chkmail score : . : R=. U=. O=. B=0.000 -> S=0.000
X-j-chkmail-Status: Ham
Archived-At: <https://mailarchive.ietf.org/arch/msg/babel/jkKfE7_UujWGY7Z8esqJ4GjQ19I>
Subject: [babel] About EIGRP [was: Thoughts on rfc6126bis]
X-BeenThere: babel@ietf.org
X-Mailman-Version: 2.1.22
Precedence: list
List-Id: "A list for discussion of the Babel Routing Protocol." <babel.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/babel>, <mailto:babel-request@ietf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/babel/>
List-Post: <mailto:babel@ietf.org>
List-Help: <mailto:babel-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/babel>, <mailto:babel-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 28 Nov 2017 16:06:22 -0000

>     At this point, the whole network must be rebooted in order to solve
>     the starvation; this is essentially what EIGRP does when it performs
>     a global synchronisation of all the routers in the network with the
>     source (the "active" phase of EIGRP).

I am open to a different formulation, but I don't really see a good way of
keeping the intent while remaining reasonably concise.  I guess I could
replace "the whole network" with "the whole starving subtree", which would
be closer to what EIGRP does.

> The active process is not a "network reboot," but rather it's very
> similar to what BABEL does.

That's an interesting debate.

[...]

> This is very similar to what BABEL does, but with two
> differences. First, there are no sequence numbers. Instead of sequence
> numbers, EIGRP assumes a route received as part of an answer to a query
> is always newer than the original route (and hence the feasibility
> condition may be reset). Second, rather than sending just the originator
> of the original route a request for a new route, EIGRP spreads the
> request out to all neighbors, so any alternative path will work.

I think that is very different from what Babel does.  In order to be able
to clear the old feasibility information, EIGRP must perform a synchronisation
of the whole sub-tree that is suffering from starvation.  It is my
understanding that all the starving routers freeze until they receive
confirmation that the whole subtree has cleared their routes.

So there's synchronisation over an unbounded number of nodes --
essentially a reboot of the starving subtree.  Whence the formulation
quoted above.

This is very different from what Babel does, since Babel is designed to
avoid synchronisation.  In Babel, a request is sent out, and a new route
becomes usable as soon as a new sequence number is received -- no need to
wait for the other neighbours to acknowledge, so a slow or crashed router
will not delay reconvergence as long as faster routers are available to
forward the sequence number.

> I think what the paragraph is trying to refer to is the (old) Stuck in
> Active (SIA) process, not the active process.

I'm not entirely clear on what SIA is, since EIGRP folks are not very keen
on discussing it.  My understanding is that the synchronisation required
by the active phase does not necessarily terminate (a router could crash
between receiving a request and sending the acknowledgment), and in this
case, the network will deadlock until the SIA timeout triggers.  But
I could be wrong.

> In more recent code, EIGRP does not reset the neighbor adjacency, but
> rather queries again (the SIA query process), and again treats any route
> received in response to this query as newer than the no longer useable
> route (because of the topology change).

So basically restarting the active phase instead of dropping the
neighbour?  There's still a timeout dependent on the slowest router in the
starving component, right?

> I would change this to -- "EIGRP marks a route active, transmitting
> a diffusing query through the network to find an alternative route in
> the case of starvation."

I think that is less informative than what I've written.  Does changing
"the whole network" to "the starving subtree", as suggested above, fix it
for you?

-- Juliusz