Re: [tsvwg] Another tunnel/VPN scenario (was RE: Reasons for WGLC/RFC asap)

Sebastian Moeller <moeller0@gmx.de> Thu, 03 December 2020 13:42 UTC

Return-Path: <moeller0@gmx.de>
X-Original-To: tsvwg@ietfa.amsl.com
Delivered-To: tsvwg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E9BF73A0B0A; Thu, 3 Dec 2020 05:42:27 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.649
X-Spam-Level:
X-Spam-Status: No, score=-1.649 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, FREEMAIL_ENVFROM_END_DIGIT=0.25, FREEMAIL_FROM=0.001, RCVD_IN_MSPIKE_H2=-0.001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001, URIBL_BLOCKED=0.001] autolearn=no autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (1024-bit key) header.d=gmx.net
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 8QLCWWeU7q4Z; Thu, 3 Dec 2020 05:42:25 -0800 (PST)
Received: from mout.gmx.net (mout.gmx.net [212.227.17.22]) (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 793863A0AA7; Thu, 3 Dec 2020 05:41:35 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=gmx.net; s=badeba3b8450; t=1607002844; bh=d+wZZHViN4mwszuM3bJb7KQZ+uwbIhlNn+sGLEOY6pk=; h=X-UI-Sender-Class:Subject:From:In-Reply-To:Date:Cc:References:To; b=b8HDRRwnuwsrSR31GO6ffb23nGOks8RXJEEPqGooclGiKacZF7vku2JHd1cjmapcQ usDy5ECH33ap4xEpbMW16o/7WYi9zMwtSsc4dRegBvG1ZYf+AV3Ni1ZDaXYA3VxLuA w4dXGhWpow+9TbaFSSUB6ja0fm51G8M9BhxFinyY=
X-UI-Sender-Class: 01bb95c1-4bf8-414a-932a-4f6e2808ef9c
Received: from [192.168.250.102] ([134.76.241.253]) by mail.gmx.com (mrgmx104 [212.227.17.168]) with ESMTPSA (Nemesis) id 1M7b6b-1koPLS380L-0084g5; Thu, 03 Dec 2020 14:40:43 +0100
Content-Type: text/plain; charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.17\))
From: Sebastian Moeller <moeller0@gmx.de>
In-Reply-To: <0763351c-3ba0-2205-59eb-89a1aa74d303@bobbriscoe.net>
Date: Thu, 3 Dec 2020 14:40:40 +0100
Cc: Ingemar Johansson S <ingemar.s.johansson=40ericsson.com@dmarc.ietf.org>, "Black, David" <David.Black@dell.com>, Pete Heist <pete@heistp.net>, tsvwg IETF list <tsvwg@ietf.org>
Content-Transfer-Encoding: quoted-printable
Message-Id: <CC0517BE-2DFC-4425-AA0A-0E5AC4873942@gmx.de>
References: <MN2PR19MB4045A76BC832A078250E436483E00@MN2PR19MB4045.namprd19.prod.outlook.com> <HE1PR0701MB2876A45ED62F1174A2462FF3C2FF0@HE1PR0701MB2876.eurprd07.prod.outlook.com> <56178FE4-E6EA-4736-B77F-8E71915A171B@gmx.de> <0763351c-3ba0-2205-59eb-89a1aa74d303@bobbriscoe.net>
To: Bob Briscoe <in@bobbriscoe.net>
X-Mailer: Apple Mail (2.3445.104.17)
X-Provags-ID: V03:K1:xvBbAJwIPc84xtssZZBCsJY8ijQ9h5XDGFls7XuZR8h9XRsV4Xv Ddca9Y73jvLtvX0dzvg4vU9re98OYPAQl7CjDTcOGFKGFnOwwjCGLjIeMvpsIe03dTY7EYE C1NuYtuLMEDQRur1VoSJJ8l5XbzEZB7RDYhqrGFB4rQ/fkqJiUV4EjnwX/a5WQpkOeUtkB8 UctS7PuX0MUYNysodB/tA==
X-UI-Out-Filterresults: notjunk:1;V03:K0:ojF7LrR+W7o=:kRunyvq8XzZ3QidYo1C81s SAVfCIDOJElT09tjh9GYcM8J1hvDXUwtKXEqg29FK4vB1lpZddIW/ZXQstCp/osIhcSRLRpNm zzae5/ajf1Y1E155GILvU0zGQ/dyluFGvZE86ZRLjgANzdkRR1Er7rpWfr2O0UaLGYqoyEiF9 py7MhB94+chKh0hA5V5UEUQuAme0wM/jD739wKtx+BTRrUM/d6pAyEwTCbjdx4KtJS0oFdnj9 ki3W1TU82oa1ODBr+9daSN7n5ZNLS1dy/IJZr/dK7+Rt1L5zDjRzaTrpm2pxfnLI2kelFeIbs 4Qe7nk4Cgrov+kTFqEKuZkQY9miTXklDo9B10GGBMAr+Rg3NUwTFwLMYDt8mM6YDk7UCRG4A4 w4oGfwujFEbhMjLxkwVv/UlGc+paQKLBbedJfNpHOD9YU7zALgHRw16n6UhUu+wZjFSsU0z0/ yykPE7mnAeq0ET/ESothaEe2i0xZhUzMpZKzucjZggfmPRRCTzI3wD2p4pwNeYDCG9HOjn03N mPfmMS8hjQrYVBOpe3+TIld6k16D3+vzw9dWMj/VfcsnaUVtAKduslVZjO0CFGdePbeQwc7k4 qKo4HXcrb2bSv/Zi6HLxYvC0wt0s0OWF9rol6YmA/fd172uue3DhP6BIXLRpr5y3gtcXeFNq/ nl6zQF3SzSeKsssd6aKMTVh8LtfvJ4yWSvK1doQ0/OfEnwehtvXZK27lBUSfiaqa02++Y48/e oiHJfcQTLDXIhrcJVWoKLkkIWbHo5lzAj5LlPUz85GSs3KcQvvPB55EEGe3m7IdCUxyK7aeEX o2PYAVOTPhWm6kFt1l2UxNEk9VGsPGryTCs9zBOn8ZafzKwl9E1Uz24Yrv7/tONnV+0NCoLu4 4p7POLdNZVxWdQU4pp1A==
Archived-At: <https://mailarchive.ietf.org/arch/msg/tsvwg/7swGAUXcVCloAwUy04p3YuyvMiM>
Subject: Re: [tsvwg] Another tunnel/VPN scenario (was RE: Reasons for WGLC/RFC asap)
X-BeenThere: tsvwg@ietf.org
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Transport Area Working Group <tsvwg.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsvwg>, <mailto:tsvwg-request@ietf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/tsvwg/>
List-Post: <mailto:tsvwg@ietf.org>
List-Help: <mailto:tsvwg-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsvwg>, <mailto:tsvwg-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 03 Dec 2020 13:42:28 -0000

Bob,

more beow in-line, prefixed [SM].

> On Dec 3, 2020, at 13:19, Bob Briscoe <in@bobbriscoe.net> wrote:
> 
> Sebastian, Jonathan,
> 
> Encrypted tunnels not revealing the component flows in the payload is a correct property of layering, which has been a principle that guides protocol design since long before the Internet protocols adopted it. I would ask you to imagine standing up (virtually) in an IETF plenary (or any networking forum) and blame layering (and/or layered encryption) for FQ not being able to isolate component flows. You would be flamed to a crisp.

	[SM] Sure, but by that same token passing of ECN codepoints between inner and outer is also a layering violation, if we did not do this, we would have no issue. But still "moving" ECN codepoints around at e- and de-tunneling is "endorsed" by the IETF. The same IETF now should make sure that these recommendations do not lead to negative outcomes, like some flows taking over most of the tunnel, by say consciously redefining the response to those ECN signals by transport protocols. 

> 
> Certainly, a flow (such as L4S) that needs to be isolated from certain other flows could be seen as the cause of the need for FQ. But it is not. It is the cause of the need for isolation, not specifically /flow/ isolation. It is FQ that chooses to rely on identifiers that violate layering in order to provide that isolation.

	[SM] Well, as long as all flows behave TCP-friendly (as they arguably should) that isolation-failure is pretty benign, no? The issue here is that L4S proposes a new way of ECN-signalling that has exciting new TCP-unfriendly failure modes, like a shared tunnel with rfc3168 flows over a rfc3168 bottleneck. IMHO, if somebody uses a tunnel, they deserve to be treated like a single flow, what is not okay, is to have some mis-designed protocols take over the tunnel for themselves.
	IMHO, looking at essential IP and TCP/UDP header fields for flow definition is the only sane way of doing it (we can haggle which to include), any field that can be gamed, will be gamed, so flows need to be defined as little game-able as possible, and if that means looking into protocol headers so be it.

> 
> The root cause of the problem can be objectively determined. FQ doesn't correctly schedule flows within encrypted tunnels, whether L4S is present or not. Therefore L4S cannot be the root cause of this problem.

	[SM] No arguable FQ is doing what it is supposed, if something presents itself to the network as a single flow it gets treated as such, if the source wanted different treatment, it knows what to do....
But again it is L4S that turns this into a problem, because in the condition we currently discuss, it fails to follow the principle of TCP-fairness... So you basically claim, if you unfairly tackle someone while the referee isn't watching that constitutes no foul. Not sure I agree with that position.

	IMHO, the root cause of this issue is still L4S hare-brained design of thinking that re-defining the meaning of CE is a safe proposition (see how I picked up your Leporidae-theme from below?). The tunnel with the rfc3168 is only the example the demonstrates that just saying anybody who cares should use an FQ-AQM is not a solution, but honestly, that is a position team L4S introduced to the discussion as a means to keep declaring the fall-out from re-defining CE "not a big deal".
Again to be clear, the problem is that L4S is designed to be actively hostile towards/over rea rfc3168 bottlenecks, and the ideal fix would be to change L4S here. there have been several proposals how to achieve that, e.g. by Jake Holland, that all have been essentially ignored by team L4S (sure such a change would requre some re-engineering, so the argument that engineering would be required is IMHO not a real blocker for Jake's proposal).

> 
> It is instructive to see how the world looks to people who are stuck so far down the FQ rabbit-hole
> that the walls of the FQ burrow and the other rabbits down there give the comforting feeling that FQ is the whole world. Except for that small chink of light in the sky when looking up out of the rabbit-hole, which is all that is visible of the rest of the universe of networking.

	[SM] Does this actually contain actionable content you want an answer to? If so please clarify, otherwise I will ignore that as an unwelcome attempt at derailing the discussion.
	I understand, that you dislike FQ (you have not been shy or subtle about that position), but the issue at hand is really an issue about L4S's proposed schemes inherent lack of safety.

Best Regards
	Sebastian



> 
> 
> Bob
> 
> On 20/11/2020 07:56, Sebastian Moeller wrote:
>> Ingemar,
>> 
>> encrypted tunnels not revealing the individual component flows in their payload is a feature of encryption and not a failure of flow isolation... Arguably an encrypted tunnel that disguises as a single flow should not allow propagation of ECN codepoints between the inner and outer layer at all, but then that is in the hand of the tunnel operator, not the AQM node.
>> 	It is quite interesting though, how tunneling is brought as an argument against the SCE proposal (only CE is guaranteed to be passed between layers*) and the very moment L4S shows issues with tunneling this is interpreted as someone else's problem. This constant application of double standards alone should be reason to reject the L4S drafts....
>> 
>> 
>> Best Regards
>> 	Sebastian
>> 
>> 
>> *) one of the original sins in regards to ECN and tunnels seems to have been not simply requiring complete unconditionally copying of inner ECN bits to outer ECN bits on en-capsulation and outer to inner on de-capsulation and letting the end-points deal with any accidental fall-out (to be complete, a tunnel should either not do any ECN propagation in any direction, or the one described). For rfc3168, I can fully understand why that route was not chosen, but years later for rfc6040 that decision is much harder to rationalize. 
>> 
>> 
>> On 20 November 2020 07:04:56 CET, Ingemar Johansson S <ingemar.s.johansson=40ericsson.com@dmarc.ietf.org> wrote:
>> Hi David, Pete
>>  
>> I try to make it clear to me what this scenario show is about and somehow I see it more as a flow isolation problem that makes FQ non-functional rather than an L4S problem?. 
>>  
>> There is of course a possibility that VPNs do not implement RFC6040 properly. I guess for software VPNs it is only an update cycle away, more hard/firmware VPNs can of course be a different story but I guess that, similar to the discussion on home gateways and ECN a few months ago, they can be upgradeable too  ?
>>  
>> /Ingemar
>>  
>>  
>> From: tsvwg <tsvwg-bounces@ietf.org> On Behalf Of Black, David
>> Sent: den 19 november 2020 22:20
>> To: Pete Heist <pete@heistp.net>
>> Cc: tsvwg IETF list <tsvwg@ietf.org>
>> Subject: [tsvwg] Another tunnel/VPN scenario (was RE: Reasons for WGLC/RFC asap)
>>  
>> [posting as an individual]
>>  
>> > I'll leave it to the WG to come up with examples of what types of tunnels and traffic scenarios could lead to this,
>> > but one example is a user who has a privacy VPN on their PC, and fq_codel on their home gateway.
>> > Let's say one flow connects to an L4S capable server, and another flow to a non-L4S, conventional server.
>> > The L4S flow will dominate the non-L4S one (whether it's ECN capable or not), probably causing some level
>> > of poor service, perhaps for a video stream, download, or whatever.
>>  
>> It’s more than home gateways – there will be increasing use of VPNs with public or shared WiFi to block snooping by other WiFi devices and/or the access point infrastructure.  In that case, the WiFi access point and nodes between the access point and the VPN gateway can only look at the outer IP header applied by the VPN.  If the VPN preserves packet boundaries and complies with RFC 6040, then ECT(1) in the inner header will show up in the outer header, but not all VPNs do both of those.
>>  
>> Thanks, --David
>>  
>> From: tsvwg <tsvwg-bounces@ietf.org> On Behalf Of Pete Heist
>> Sent: Thursday, November 19, 2020 3:03 PM
>> To: Gorry Fairhurst
>> Cc: tsvwg IETF list
>> Subject: Re: [tsvwg] Reasons for WGLC/RFC asap
>>  
>> [EXTERNAL EMAIL] 
>> 
>> On Thu, 2020-11-19 at 16:34 +0000, Gorry Fairhurst wrote:
>> On 19/11/2020 16:22, Pete Heist wrote:
>>  
>> Hi Koen,
>>  
>> Rather than thinking of this as advantages and disadvantages to waiting, I see it as an engineering process. It was decided earlier this year that the L4S proposal has enough support to continue, so we're on that path now. Part of that decision, as I understood it, also recognized that there are valid safety concerns around compatibility with existing AQMs, and some solution needs to be devised.
>>  
>> RFC3168 bottleneck detection was added to TCP Prague, which appears to be difficult to do reliably when there is jitter or cross-flow traffic, and it has since been disabled in the reference implementation. The l4s-ops draft was started, but isn't complete yet and may need WG adoption as part of a LC. We can then decide how effective the proposed mitigations are against the risks and prevalence.
>>  
>> To start a WGLC now would circumvent that earlier recognition that a safety case needs to be made. Meanwhile, since testing showed that tunnels through RFC3168 FQ AQMs are a straightforward path to unsafe flow interaction, along with other issues relative to the goals, it doesn't seem like the engineering process is done just yet.
>>  
>> By the way, I liked your data - and it helped me a lot to look at this, thanks very much for doing this.
>> 
>> I'm glad, as I think we're at our best when we're doing engineering and producing data. I wish it were easier to do!
>> It would help me if you clarify what you mean by  "unsafe" - to me "safety" relates to traffic unresponsive to drop, as in CBR traffic, etc. I've not understood how CE-marked traffic can erode safety, but maybe I missed something?
>> 
>> Sure, so the existing RFC3168 CE signal in use on the Internet today indicates an MD (multiplicative decrease), whereas the redefined CE signal in L4S indicates an AD (additive decrease). Two congestion controls responding to CE in a different way, or one that responds to CE with an AD and one that responds only to drop (i.e. all standard congestion controls that advertise Not-ECT), will not interact safely in the same RFC3168 signaling queue. We're probably on the same page here already, but I'll refer to section 5 of RFC8257.
>>  
>> That is one of the reasons why ECT(1) is used in L4S to place L4S flows in the L queue- to keep them separate from conventional flows in the C queue. As long as flows have advertised their capability correctly, that works.
>>  
>> However, existing RFC3168 queues do not have knowledge of L4S, therefore will not know that ECT(1) means that traffic needs to be segregated and signaled in a different way. They will signal a Prague flow, which sets ECT(1), with CE, expecting the flow to respond with an MD, rather than AD. Meanwhile they'll signal an RFC3168 or non-ECN flow with either CE or drop, and in either case the flow will respond with an MD, causing conventional flows to yield to Prague flows to varying degrees depending on the AQM in use.
>>  
>> Here's an example of CUBIC and Prague when they end up in the same fq_codel queue:
>> http://sce.dnsmgr.net/results/l4s-2020-11-11T120000-final/l4s-s6-rfc3168-1q/l4s-s6-rfc3168-1q/l4s-s6-rfc3168-1q-ns-prague-vs-cubic-fq_codel-50Mbit-20ms_tcp_delivery_with_rtt.svg
>>  
>> Here's a more extreme example of Reno and Prague sharing a single PIE queue with ECN enabled (less common):
>> http://sce.dnsmgr.net/results/l4s-2020-11-11T120000-final/l4s-s6-rfc3168-1q/l4s-s6-rfc3168-1q/l4s-s6-rfc3168-1q-ns-prague-vs-reno-pie-50Mbit-20ms_tcp_delivery_with_rtt.svg
>>  
>> In the example with PIE, Reno appears to be driven at or close to minimum cwnd. In the fq_codel example, the steady state throughput of Prague:CUBIC is around 19:1. We've seen a range in the Codel case from around 12:1 to 20:1. In my opinion, we could use the word "unsafe" here in both cases.
>> I'm not sure why "tunnels have crept in here. There have always been side-effects with classification (and hence scheduling), but I don't see new issues relating to "tunnels" with ECN.
>> 
>> Tunnels are relevant because they provide an easy practical path to the unsafe flow interaction described above. The widely used fq_codel qdisc has ECN enabled by default. Fortunately, because it has flow-fair queueing, Prague flows and conventional flows are usually placed in a separate queue (hash collisions aside), causing Prague to only affect itself with additional delay (TCP RTT). However, a tunnel's encapsulated packets all share the same fq_codel queue because they all have the same 5-tuple, so there is unsafe interaction between the tunnel's flows. Here we use Wireguard through fq_codel:
>>  
>> http://sce.dnsmgr.net/results/l4s-2020-11-11T120000-final/l4s-s5-tunnel/l4s-s5-tunnel-phys-wireguard-prague-vs-cubic-fq_codel-50Mbit-20ms_tcp_delivery_with_rtt.svg
>>  
>> I'll leave it to the WG to come up with examples of what types of tunnels and traffic scenarios could lead to this, but one example is a user who has a privacy VPN on their PC, and fq_codel on their home gateway. Let's say one flow connects to an L4S capable server, and another flow to a non-L4S, conventional server. The L4S flow will dominate the non-L4S one (whether it's ECN capable or not), probably causing some level of poor service, perhaps for a video stream, download, or whatever.
>> I'm not commenting on when the Chairs think a WGLC will provide useful information, we'll say that in due course.
>> 
>> Ok, I trust that we'll engage enough disinterested people into congestion control who will add their input.
>>  
>> Thanks Gorry for looking this over. :)
>> Best wishes,
>> 
>> Gorry
>> 
>>  
>> Regards,
>> Pete
>>  
>> On Wed, 2020-11-18 at 10:31 +0000, De Schepper, Koen (Nokia - BE/Antwerp) wrote:
>>  
>> Hi all,
>>  
>> To continue on the discussions in the meeting, a recap and some extra thoughts. Did I miss some arguments?
>>  
>> Benefits to go for WGLC/RFC asap:
>> 	• There is NOW a big need for solutions that can support Low Latency for new Interactive applications
>> 	• The big L4S benefits were a good reason to justify the extra network effort to finally implement ECN in general and AQMs in network equipment
>> 	• Timing is optimal now: implementations in NW equipment are coming and deployment can start now
>> 	• Deployment of L4S support will include deployment of Classic ECN too! So even for the skeptics among us, that consider that the experiment can fail due to CCs not performing to expectations, we will fall back to having Classic ECN support
>> 	• Current drafts are about the network part, and are ready and stable for a very long time now.
>> 	• Only dependency to CCs in the drafts are the mandatory Prague requirements (only required input/review from future CC developers: are they feasible for you)
>> 	• We have a good baseline for a CC (upstreaming to Linux is blocked by the non-RFC status)
>> 	• Larger scale (outside the lab) experiments are blocked by non-RFCs status
>> 	• It will create the required traction within the CC community to come up with improvements (if needed at all for the applications that would benefit from it; applications that don’t benefit from it yet, can/will not use it)
>> 	• NW operators have benefits now (classic ECN and good AQMs) and in the future can offer their customers better Low Latency experience for the popular interactive applications
>> 	• When more L4S CCs are developed, the real independent evaluation of those can start
>>  
>> Disadvantages to wait for WGLC/RFC:
>> 	• We’ll gets stuck in an analysis paralysis (aren’t we already?)
>> 	• Trust in L4S will vanish
>> 	• No signs that we can expect more traction in CC development; trust and expectations of continuous delays will not attract people working on it, as there will be plenty of time before deployments are materializing
>> 	• Product development of L4S will stall and die due to uncertainty on if L4S will finally materialize
>> 	• Product development of Classic ECN will stall and die due to uncertainty on how L4S will finally materialize
>>  
>> What are the advantages to wait? Do they overcome these disadvantages?
>>  
>> Regards,
>> Koen.
>>  
>>  
>>  
>> 
>> -->
> 
> -- 
> ________________________________________________________________
> Bob Briscoe                               
> http://bobbriscoe.net/
> 
>                PRIVILEGED AND CONFIDENTIAL
>