From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-ig0-x236.google.com (mail-ig0-x236.google.com [IPv6:2607:f8b0:4001:c05::236]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (Client CN "smtp.gmail.com", Issuer "Google Internet Authority G2" (verified OK)) by huchra.bufferbloat.net (Postfix) with ESMTPS id 9B94A21F325 for ; Thu, 6 Nov 2014 09:47:29 -0800 (PST) Received: by mail-ig0-f182.google.com with SMTP id hn18so3832168igb.15 for ; Thu, 06 Nov 2014 09:47:29 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:date:message-id:subject:from:to:content-type; bh=h3VkvkXGt0mdcJ1D/I4G6FBkgpbtSHL7Wu6DFw3wvPs=; b=kfPL6riF5NZ8r49ce5pfPeWfRmOkUJZw/XJll33WQt3/BrTtTULY1TABk8EFFhx5mL niB+3BxrUJ6t+XK6BeV2s8BsR+Io4drrfqcLpZ0qwFREtDus1ZK9kH+wq4BXAlA7YYYY NtaRAvfWKP3xjw8cmULvblaxU2IFnHE2PIcNX4Sw9DvqYYx+t9wHci0bdobVsSRTGIQ5 oop1za5524NGVAbST5pnMKza1vIH/P2iU78uD0MuiDuyU27lJtPocPkK5xI7nDOCO7St 4JDZmx4swkvp8AY9pZZyQUgEjs/ARokRdxFImnal87BN6xape6II0ZGl9lM3EZE+A+AB mG+Q== MIME-Version: 1.0 X-Received: by 10.51.16.99 with SMTP id fv3mr41912079igd.31.1415296048818; Thu, 06 Nov 2014 09:47:28 -0800 (PST) Received: by 10.64.98.136 with HTTP; Thu, 6 Nov 2014 09:47:28 -0800 (PST) Date: Thu, 6 Nov 2014 12:47:28 -0500 Message-ID: From: Justin McCann To: bloat Content-Type: multipart/alternative; boundary=001a1134a5223df20f0507344a39 Subject: [Bloat] ToS bits being set to mitigate transit congestion problems X-BeenThere: bloat@lists.bufferbloat.net X-Mailman-Version: 2.1.13 Precedence: list List-Id: General list for discussing Bufferbloat List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 06 Nov 2014 17:47:58 -0000 --001a1134a5223df20f0507344a39 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable The m-labs discussion list has had some traffic lately trying to figure out why NDT throughput changed abruptly. Apparently Cogent (and others?) has started setting ToS bits on retail customer traffic, and leaving wholesale traffic as-is (see below). https://groups.google.com/a/measurementlab.net/forum/#!topic/discuss/vcQnaZ= JO6nQ Observed ToS settings include 0x00, 0x28, and 0x48: https://groups.google.com/a/measurementlab.net/forum/#!topic/discuss/ec_-U5= nae5E Nice to see at least some network operators talking about what they're doing. Justin
Due to the severe level of congestion, the lack of movement in negotiating possible remedies and the extreme level of impact to small enterprise customers (retail customers), Cogent implemented a QoS structure that impacts interconnections during the time they are congested in February and March of 2014. Consistent with recommendations from BITAG (Broadband Internet Technical Advisory Group: http://www.bitag.org/documents/BITAG_-_Congestion_Management_Report.pdf), Cogent prioritized based on user type putting its retail customers in one group and wholesale in another. Retail customers were favored because they tend to use applications, such as VoIP, that are most sensitive to congestion. M-Labs is set up in Cogent=E2=80=99s system as a retail custom= er and their traffic was marked and handled exactly the same as all other retail customers. Additionally, all wholesale customers traffic was marked and handled the same way as other wholesale customers. This was a last resort effort to help manage the congestion and its impact to our customers. Hank Kilmer Cogent
--001a1134a5223df20f0507344a39 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable

The m-labs discussion list has had some traffic lately tryi= ng to figure out why NDT throughput changed abruptly.

Apparently Cogent (and others?) has st= arted setting ToS bits on retail customer traffic, and leaving wholesale tr= affic as-is (see below).=C2=A0

Observed ToS settings include 0x00, 0x28,= and 0x48:
Nice to see at least some network operators talking about what = they're doing.

=C2=A0 Justin

<blockquote>
Due to the severe level of congestion,= the lack of movement in negotiating possible remedies and the extreme leve= l of impact to small enterprise customers (retail customers), Cogent implem= ented a QoS structure that impacts interconnections during the time they ar= e congested in February and March of 2014.=C2=A0 Consistent with recommenda= tions from BITAG (Broadband Internet Technical Advisory Group:=C2=A0= http://www.bitag.org/documents/BITAG_-_Congestio= n_Management_Report.pdf), Cogent prioritized based on user type putting its retail customer= s in one group and wholesale in another.=C2=A0 Retail customers were favore= d because they tend to use applications, such as VoIP, that are most sensit= ive to congestion.=C2=A0 M-Labs is set up in Cogent=E2=80=99s system as a r= etail customer and their traffic was marked and handled exactly the same as= all other retail customers.=C2=A0 Additionally, all wholesale customers tr= affic was marked and handled the same way as other wholesale customers.=C2= =A0 This was a last resort effort to help manage the congestion and its imp= act to our customers.=C2=A0

Hank Kilmer=C2=A0
Cogent=C2=A0
</blockquote>
--001a1134a5223df20f0507344a39--