From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-qk1-x741.google.com (mail-qk1-x741.google.com [IPv6:2607:f8b0:4864:20::741]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id A6F613B29E for ; Thu, 1 Nov 2018 15:13:34 -0400 (EDT) Received: by mail-qk1-x741.google.com with SMTP id r71so13014878qkr.10 for ; Thu, 01 Nov 2018 12:13:34 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=mime-version:references:in-reply-to:from:date:message-id:subject:to :cc:content-transfer-encoding; bh=x3DeyG7m7Qez7jhe3XK8pYKH940nUPDhq6BH94mpXw4=; b=eDkqDMjpK1f7r85TDBMLYEWCEEXtQmiNRrLgO03vgNPtcYuuL632pLt4wWiCVHbNld MGTOASUT0BVHtTp9cwlCyfov7qd/2pUAuWfWWFuDDp8nSr0eQewhqErDgnPoI+qVWgN+ dgCRnBAPpS3ashqVoRBdLpfYKMQh+je+ZGckorE+QXqHDlY/qmov5v5buiZ8caN8lC+9 uqaz+rPju/fRpuzwDfe+GRUx4/S6UcSxtHC8UvUz/qzFoG6YO/dg31JSe0vf0gck7HPE Uc6Gn0YaZx1/QgP9HbV3zELGe1+vmIDnzYzZwuuObhdXwsZBwe+BEGAWwCYVOW4J2VF4 TkeA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc:content-transfer-encoding; bh=x3DeyG7m7Qez7jhe3XK8pYKH940nUPDhq6BH94mpXw4=; b=NogQVago4SMlrQsVAQgxj5tY161YHe286KLMAtTQgBCMF0DIdjS/IZ13vp/DWM2kI1 oXuAEzH2DCIxAfa9ZO/P5IKpdajZ15MDazYIPkoKk8sSERvQHzhhpIJw03G8HP6VedJL drJai6tHXweiCOrk0kGEutqCQ6u7cKhNZknFIjpRsUktXUhO3qBDAR3Ht6bpuSIqy5ba MZ0b7lgmjzfm4CVmwgReOwwospAgNyfOeuCw4+LikAlXe/b3eppDv62weFHhHE1SZF1I uKL8I2ejmNA8ZDlIh86r/04JqwhTQegn/JOk0IjztSSTshNqGj9UKHgOmoi+sOyQqjlO wHeg== X-Gm-Message-State: AGRZ1gL0pBBIcpf2eNJudtBgBJ3gfWbskn/FEqVUTvq3cOuSYfq/jh54 yMKEMNoLbgy3Jf0DdE+3VpH65SMuNoRDigSGcaQ= X-Google-Smtp-Source: AJdET5dl2F5BJsJp7sIieMmZ7hFgy6zylwL+nMWJOiK4g1nH62Nhh04WsxRqDHNt0zHLPJrsU4RO2WYn1X4UCSpYezc= X-Received: by 2002:ac8:5314:: with SMTP id t20mr7801913qtn.328.1541099614039; Thu, 01 Nov 2018 12:13:34 -0700 (PDT) MIME-Version: 1.0 References: <878t2h1jtm.fsf@taht.net> <877ei096vi.fsf@toke.dk> <4F59C958-0AF9-4531-B700-0A64572E22CF@cablelabs.com> <874ld1q6aa.fsf@toke.dk> In-Reply-To: <874ld1q6aa.fsf@toke.dk> From: Dave Taht Date: Thu, 1 Nov 2018 12:13:19 -0700 Message-ID: To: =?UTF-8?B?VG9rZSBIw7hpbGFuZC1Kw7hyZ2Vuc2Vu?= Cc: Greg White , =?UTF-8?Q?Dave_T=C3=A4ht?= , tsvwg@ietf.org, bloat Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Subject: Re: [Bloat] quick review and rant of "Identifying and Handling Non Queue Building Flows in a Bottleneck Link" X-BeenThere: bloat@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: General list for discussing Bufferbloat List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 01 Nov 2018 19:13:34 -0000 On Thu, Nov 1, 2018 at 6:25 AM Toke H=C3=B8iland-J=C3=B8rgensen wrote: > > Greg White writes: > > > Hi Toke, thanks for the pointer to your paper, I had not seen it > > before. > > You're welcome :) > > > I agree that could be a candidate algorithm. It is certainly simple. > > It may not be the only (or perhaps even the best) solution for the > > dual-queue case though. I'm thinking in the context of an L4S TCP > > flow, which can respond quickly to "new" ECN markings and achieve link > > saturation with ultra low (but not zero) queuing delay. A good > > property for a queue protection algorithm would be that these L4S > > flows could be consistently placed in the NQB queue. I think that the > > simple approach you mentioned would result in many L4S flows being > > deemed Queue Building. > > Yes, I think you are right (depending on traffic mix, of course). It > might be possible to tweak it to work better, though. E.g., by changing > the threshold (moving flows to QB if they end up with more than X ms of > queue). This would only work if you start out all flows at NQB, with the > associated aggressive marking behaviour; so I'm not sure if a normal TCP > flow would ever manage to get to QB state before getting clobbered by > the NQB markings... > > > Additionally, we've observed applications that send variable sized > > "messages" at a fixed rate (e.g. 20 messages/second) where the message > > sometimes exceeds the MTU and results in two closely spaced (possibly > > back-to-back) packets. This is a flow that I think should be > > considered to be NQB, but would get flagged as QB by the simple > > approach. You described this case in your paper, where you state that > > the first Q bytes of each burst will be treated as NQB (the first > > packet in the case we're talking about here), but the rest will be > > treated as QB. Assuming that message latency is important for these > > sorts of applications, this is equivalent to saying that the entire > > burst is considered as QB. In the fq_codel case, the message latency > > would be something like Q(n-1)(N+1)/R (assuming no other sparse flow > > arrivals), something like 1.3ms using the example values in your paper > > (plus n=3D2, N=3D10) which may be ok. In the dual-queue case it is a > > bigger deal, because the remaining packets would be put at the end of > > the QB queue, which could have a latency of 10 or 20 ms. > > Sure, it's by no means a perfect mechanism. But it makes up for that by > it's simplicity, IMO. And it does work really well for *a lot* of > today's latency-sensitive traffic. > > (In your case of two-MTU messages, you could tune the quantum to allow > those; but of course you can construct examples that won't work). sch_fq has a default quantum of 2 mtu, with a initial burst of 10. There's all sorts of interesting work inside that to "right-size" the ongoing gso offloads and a major new advance over there on calculating rtts properly is described here: https://lwn.net/Articles/766564/ ... there was at one point, a fq_pie implementation that used the rbtree in sch_fq to achieve perfect fairness. we often tune fq_codel's quantum as low as 300, at low rates. > > So, a queue protection function that provides a bit more (but still > > limited) allowance for a flow to have packets in queue would likely > > work better in the dual-queue case. For inbound shaping sch_cake defaults to 2mtu at higher rates. This kind of opens a question in that, what is a typical target bandwidth for l4s applications? the videoconferencing paper I dissed in my earlier rant focused only at 2mbits (and drew the conclusion that sfq was the best option for the cc algo's 300ms desired range) I have generally been focused on badwidths in the 4mbit-40gbit range. > > Yeah, that's tricky, especially if you want it to be very accurate in > its distinction; which I sort of gather that you do, right? In part, that is why I would like the language in the problem statement clarified. To give a concrete counterexample, think upon the ultimate choice of a crc32 algorithm and the parameters that drove that choice. If an accuracy of identifying "sparse flows" or NQB flows can be specified, then all sorts of algorithms can be thrown into the fray. AI. Bloom filters. rbtrees. regexes. cookoo++ hashes... and boring old fq tech. :) > -Toke > _______________________________________________ > Bloat mailing list > Bloat@lists.bufferbloat.net > https://lists.bufferbloat.net/listinfo/bloat --=20 Dave T=C3=A4ht CTO, TekLibre, LLC http://www.teklibre.com Tel: 1-831-205-9740