From mboxrd@z Thu Jan 1 00:00:00 1970 From: "John A. Sullivan III" Subject: Re: Latency difference between fifo and pfifo_fast Date: Tue, 06 Dec 2011 03:39:56 -0500 (EST) Message-ID: <03369d06-ee52-4eef-a8b6-5a21e6d56840@jasiiieee> References: <1323152942.2467.16.camel@edumazet-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: netdev@vger.kernel.org To: Eric Dumazet Return-path: Received: from mout.perfora.net ([74.208.4.194]:64288 "EHLO mout.perfora.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753121Ab1LFHjf convert rfc822-to-8bit (ORCPT ); Tue, 6 Dec 2011 02:39:35 -0500 In-Reply-To: <1323152942.2467.16.camel@edumazet-laptop> Sender: netdev-owner@vger.kernel.org List-ID: ----- Original Message ----- > From: "Eric Dumazet" > To: "John A. Sullivan III" > Cc: netdev@vger.kernel.org > Sent: Tuesday, December 6, 2011 1:29:02 AM > Subject: Re: Latency difference between fifo and pfifo_fast >=20 > Le mardi 06 d=C3=A9cembre 2011 =C3=A0 07:02 +0100, Eric Dumazet a =C3= =A9crit : > > Le lundi 05 d=C3=A9cembre 2011 =C3=A0 23:10 -0500, John A. Sullivan= III a > > =C3=A9crit : > > > Hello, all. We are trying to minimize latency on our iSCSI SAN. > > > The > > > network is entirely dedicated to the iSCSI traffic. Since all > > > the > > > traffic is the same, would it make sense to change the qdisc for > > > that > > > interface to fifo from the default pfifo_fast or is the latency > > > difference between the two completely negligible? Thanks - John > >=20 > > A very small difference indeed. How many packets per second are > > expected ? What kind of NIC are you using ? > >=20 >=20 > To really remove a possible source of latency, you could remove qdisc > layer... >=20 > ifconfig eth2 txqueuelen 0 > tc qdisc add dev eth2 root pfifo > tc qdisc del dev eth2 root >=20 >=20 >=20 >=20 Really? I didn't know one could do that. Thanks. However, with no que= ue length, do I have a significant risk of dropping packets? To answer = your other response's question, these are Intel quad port e1000 cards. = We are frequently pushing them to near line speed so 1,000,000,000 / 1= 534 / 8 =3D 81,486 pps - John