Date: Tue, 21 May 2013 07:52:24 -0700 (PDT) From: Barney Cordoba <barney_cordoba@yahoo.com> To: Hooman Fazaeli <hoomanfazaeli@gmail.com>, Luigi Rizzo <rizzo@iet.unipi.it> Cc: freebsd-net@freebsd.org Subject: Re: netmap bridge can tranmit big packet in line rate ? Message-ID: <1369147944.27968.YahooMailClassic@web121606.mail.ne1.yahoo.com> In-Reply-To: <20130521143906.GA80993@onelab2.iet.unipi.it>
next in thread | previous in thread | raw e-mail | index | archive | help
=0A=0A--- On Tue, 5/21/13, Luigi Rizzo <rizzo@iet.unipi.it> wrote:=0A=0A> F= rom: Luigi Rizzo <rizzo@iet.unipi.it>=0A> Subject: Re: netmap bridge can tr= anmit big packet in line rate ?=0A> To: "Hooman Fazaeli" <hoomanfazaeli@gma= il.com>=0A> Cc: freebsd-net@freebsd.org=0A> Date: Tuesday, May 21, 2013, 10= :39 AM=0A> On Tue, May 21, 2013 at 06:51:12PM=0A> +0430, Hooman Fazaeli wro= te:=0A> > On 5/21/2013 5:10 PM, Barney Cordoba wrote:=0A> > >=0A> > > --- O= n Tue, 5/21/13, liujie <liujie@263.net>=0A> wrote:=0A> > >=0A> > >> From: l= iujie <liujie@263.net>=0A> > >> Subject: Re: netmap bridge can tranmit big= =0A> packet in line rate ?=0A> > >> To: freebsd-net@freebsd.org=0A> > >> Da= te: Tuesday, May 21, 2013, 5:25 AM=0A> > >> Hi, Prof.Luigi RIZZO=0A> > >>= =A0 =A0 =0A> > >>=A0 Firstly i should thank you for netmap. I=0A> tried to = send a=0A> > >> e-mail to you=0A> > >> yestoday, but it was rejected.=0A> >= >>=0A> > >>=A0 I used two machines to test netmap=0A> bridge. all with i7-= 2600=0A> > >> cpu and intel=0A> > >> 82599 dual-interfaces card.=0A> > >>= =0A> > >>=A0 One worked as sender and receiver with=0A> pkt-gen, the other= =0A> > >> worked as bridge=0A> > >> with bridge.c.=0A> > >>=0A> > >>=A0 as = you said,I feeled comfous too when i=0A> saw the big packet=0A> > >> perfor= mance=0A> > >> dropped, i tried to change the memory=0A> parameters of=0A> = > >> netmap(netmap_mem1.c=0A> > >> netmap_mem2.c),but it seemed that=A0 can= =0A> not resove the=0A> > >> problem.=0A> > >>=A0=A0=A060-byte packet send = 14882289=0A> pps=A0 recv =0A> > >> 13994753 pps=0A> > >>=A0=A0=A0124-byte= =A0=0A> =A0=A0=A0=0A> > >>=A0 =A0 send=A0=A0=A08445770 pps=0A> =0A> > >> re= cv=A0 =A0 7628942 pps=0A> > >>=A0=A0=A0252-byte=A0=0A> =A0=A0=A0=0A> > >>= =A0 =A0 send=A0=A0=A04529819 pps=0A> =0A> > >> recv=A0 =A0=A0=A03757843 pps= =0A> > >>=A0=A0=A0508-byte=A0=0A> =A0=A0=A0=0A> > >>=A0 =A0 send=A0 =A0 235= 0815 pps =0A> > >> recv=A0 =A0 1645647 pps=0A> > >>=A0=A0=A01514-byte=A0 = =A0=0A> =A0=A0=A0send =0A> > >>=A0=A0=A0814288 pps=A0=0A> =A0=A0=A0recv=A0 = 489133=0A> > >> pps=0A> > > These numbers indicate you're tx'ing 7.2Gb/s wi= th=0A> 60 byte packets and=0A> > > 9.8Gb/s with 1514, so maybe you just nee= d a new=0A> calculator?=0A> > >=0A> > > BC=0A> > > ________________________= _______________________=0A> > >=0A> > AsBarney pointed outalready, your num= bers are=0A> reasonable. You have almost saturated=0A> > the link with 1514= byte packets.In the case of 64 byte=0A> packets, you do not achieve line= =0A> > rate probably because of the congestion on the bus.Can=0A> you show = us "top -SI" output on the=0A> > sender machine?=0A> =0A> the OP is comment= ing that on the receive side he is seeing a=0A> much=0A> lower number than = on the tx side (A:ix1 489Kpps vs A:ix0=0A> 814Kpps).=0A> =0A> =A0 =A0 [pkt-= gen -f tx ix0]-->--[ix0 bridge ]=0A> =A0 =A0 [=A0=A0=A0HOST A=A0 =A0 =A0=0A= > =A0 ]=A0 =A0=A0=A0[=A0 =A0 HOST B ]=0A> =A0 =A0 [pkt-gen -f rx ix1]--<--[= ix1=A0 =A0=0A> =A0 =A0 ]=0A> =0A> What is unclear is where the loss occurs.= =0A> =0A> =A0=A0=A0 cheers=0A> =A0=A0=A0 luigi=0A=0AThe ixgbe driver has ma= c stats that will answer that. Just look at the=0Asysctl output.=0A=0ABC
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1369147944.27968.YahooMailClassic>