Hi, Would anybody pleassuggesme how to change internal netem queue length. I try to usto pfifo bui think it creates its own queue whilnetequeue is limited to 1000. i want to increase queue length as i havmany packeloss. Also i would i like to know whether netem does any scheduling? Thank you, Mit. Frolaurent.mariat thomson.net Thu Mar 13 08:30:03 2008 From: laurent.mariathomson.net (Marie Laurent) Date: Thu, 13 Mar 2008 16:30:03 +0100 Subject: NeteQueuLength Message-ID: <E55CA8ECFBA85E4CB8E8173566B9EFEB931021@xxxxxxxxxxxxxxxxxxxxxxxxxx> You should usthlimit keyword. If you wanto resizthe queue to 16000 packets, insert limit 16000 into your command line. Yours sincerely, LaurenMARIE -----Original Message----- From: netem-bounces alists.linux-foundation.org [mailto:netem-bounces alists.linux-foundation.org] On Behalf Of Mittal Patel Sent: jeudi 13 mars 2008 12:14 To: netealists.linux-foundation.org Subject: NeteQueuLength Hi, Would anybody pleassuggesme how to change internal netem queue length. I try to usto pfifo bui think it creates its own queue whilnetequeue is limited to 1000. i want to increase queue length as i havmany packeloss. Also i would i like to know whether netem does any scheduling? Thank you, Mit. _______________________________________________ Netemailing list Netealists.linux-foundation.org https://lists.linux-foundation.org/mailman/listinfo/netem Frotmt2004 agmail.com Thu Mar 20 15:13:48 2008 From: tmt2004 agmail.co(Tri Minh Tran) Date: Thu, 20 Mar 2008 18:13:48 -0400 Subject: latency problem Message-ID: <8e5cce0e0803201513o61226109nbbd6ec5636bf36a4@xxxxxxxxxxxxxx> Hello, I havto do an experimenthat need to measure the transmission time of aamounof data. I write the program in Java using javasocket and usneteto simulate a WAN network by setting the bandwidth and delay (latency). I measurthtransmission time as follows: Beforsending an data object, thclient records the current system tim(start) using System.currentTimeMillis() and saves into thdata object. Whethserver receives the data object, it gets the current systetimagain (end). The difference between (end-start) is the transmissiotime. Thprobleis that the DELAY (latency) set by netem seems not to be right. For example, if I sei5ms in netem, the measured latency valuis abou20ms, if I set it 10ms in netem, the measured value is 40ms, if iis 20ms by netem, thmeasured value is 56ms. I awondering if anyonhas had the same problem and if you could givmsome suggestion. I'd really appreciate it. Thank you. Tri Frotmt2004 agmail.com Fri Mar 21 07:44:59 2008 From: tmt2004 agmail.co(Tri Minh Tran) Date: Fri, 21 Mar 2008 10:44:59 -0400 Subject: ratcontrol Message-ID: <8e5cce0e0803210744u7029cf78m7a1808484bb40fd6@xxxxxxxxxxxxxx> hello, I gothis examplfrom the netem page to change the transmission rate: # tc qdisc add dev eth0 roohandl1:0 netem delay 100ms # tc qdisc add dev eth0 paren1:1 handl10: tbf rate 256kbit buffer 1600 limi3000 # tc -s qdisc ls dev eth0 qdisc nete1: limi1000 delay 100.0ms Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) qdisc tbf 10: rat256Kbiburst 1599b lat 26.6ms Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) I'trying to se'tbf rate' value to different values such as 64kbit, 128kbit, 256kbiand us'ping' to test it. However I didn't see any differencin thround trip times of sending packets. Could anyontell mif I'm doing anything wrong here? is this the way to changthtransmission rate? is this the way to test the transmissioratusing 'ping'? Thanks! From.nambiar atcs.com Mon Mar 24 23:46:30 2008 From: m.nambiar atcs.co(M Nambiar) Date: Tue, 25 Mar 2008 12:16:30 +0530 Subject: ratcontrol In-Reply-To: <8e5cce0e0803210744u7029cf78m7a1808484bb40fd6@xxxxxxxxxxxxxx> Message-ID: <OFFE6327A4.FFEC756F-ON65257417.0021EC54-65257417.00253924@xxxxxxx> Hi, EveI afacing a similar problem ..... esp for low bandwidth settings. I arunning a tesof latency with rate limitation (bandwidth). I have a simplmodel to calculatping response time, which I am trying to verify with my tests. My primitivmodel expects thping response time to be equal to two times thsuof transmission time plus latency, ignoring other effects. Iother words ping responstime = 2*(Size/ABW + Latency) where ABW = availablbandwidth. So for a ping sizof 1400 bytes and 100 ms latency and 64 kbps bandwidth I expecping responstime to be 2*(175 + 100) = 550ms. However I anoquite getting those results at all time when I use netem iconjunction with tbf/htb. I used thsamcommands as Tri did except that I used 64kbit rate instead of 256 kbiin tbf. I goa ping response time of only 337 ms. Using htb also does noimprovthe results. Wondering where I am going wrong. Thtimer accuracy is on thnetem host is 10 ms. Manoj KarunakaraNambiar Tata Consultancy Services Ph:- 67793760 Buzz:- 4223760 Cell:- 9833111377 Mailto: m.nambiar atcs.com Website: http://www.tcs.com ____________________________________________ Experienccertainty. IT Services Business Solutions Outsourcing ____________________________________________ "Tri Minh Tran" <tmt2004 agmail.com> Senby: netem-bounces alists.linux-foundation.org 03/21/2008 08:14 PM To netealists.linux-foundation.org cc Subject ratcontrol hello, I gothis examplfrom the netem page to change the transmission rate: # tc qdisc add dev eth0 roohandl1:0 netem delay 100ms # tc qdisc add dev eth0 paren1:1 handl10: tbf rate 256kbit buffer 1600 limi3000 # tc -s qdisc ls dev eth0 qdisc nete1: limi1000 delay 100.0ms Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) qdisc tbf 10: rat256Kbiburst 1599b lat 26.6ms Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) I'trying to se'tbf rate' value to different values such as 64kbit, 128kbit, 256kbiand us'ping' to test it. However I didn't see any differencin thround trip times of sending packets. Could anyontell mif I'm doing anything wrong here? is this the way to changthtransmission rate? is this the way to test the transmissioratusing 'ping'? Thanks! _______________________________________________ Netemailing list Netealists.linux-foundation.org https://lists.linux-foundation.org/mailman/listinfo/netem ForwardSourceID:NT0001821A =====-----=====-----===== Notice: Thinformation contained in this e-mail messagand/or attachments to imay contain confidential or privileged information. If you ar nothintended recipient, any dissemination, use, review, distribution, printing or copying of th informatiocontained in this e-mail messag and/or attachments to iarstrictly prohibited. If you havreceived this communication in error, pleasnotify us by reply e-mail or telephonand immediately and permanently deletthmessage and any attachments. Thank you -------------- nexpar-------------- AHTML attachmenwas scrubbed... URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20080325/f5848c76/attachment.ht Frohello.martin acomhem.se Wed Mar 26 00:05:26 2008 From: hello.martiacomhem.se (Martin Andersson) Date: Wed, 26 Mar 2008 08:05:26 +0100 Subject: ratcontrol In-Reply-To: <OFFE6327A4.FFEC756F-ON65257417.0021EC54-65257417.00253924@xxxxxxx> References: <OFFE6327A4.FFEC756F-ON65257417.0021EC54-65257417.00253924@xxxxxxx> Message-ID: <47E9F5B6.40800@xxxxxxxxx> M Nambiar wrote: > > > Iother words ping responstime = 2*(Size/ABW + Latency) where ABW = > availablbandwidth. > > So for a ping sizof 1400 bytes and 100 ms latency and 64 kbps > bandwidth I expecping responstime to be 2*(175 + 100) = 550ms. > > However I anoquite getting those results at all time when I use > netein conjunction with tbf/htb. > > I used thsamcommands as Tri did except that I used 64kbit rate > instead of 256 kbiin tbf. I goa ping response time of only 337 ms. Aryou doing thnetem command in a route/bridge in between, or is't in thmachinyou're pinging to/from. Sincneteoperate on one path only (175+100=275ms) From.nambiar atcs.com Wed Mar 26 00:21:17 2008 From: m.nambiar atcs.co(M Nambiar) Date: Wed, 26 Mar 2008 12:51:17 +0530 Subject: ratcontrol In-Reply-To: <47E9F5B6.40800@xxxxxxxxx> Message-ID: <OF225F8818.1EE83BB4-ON65257418.00275BB6-65257418.00286785@xxxxxxx> neteis operating in both paths - routing is symmetric. So thping requesand ping responsshould be affected. Manoj KarunakaraNambiar Tata Consultancy Services Ph:- 67793760 Buzz:- 4223760 Cell:- 9833111377 Mailto: m.nambiar atcs.com Website: http://www.tcs.com ____________________________________________ Experienccertainty. IT Services Business Solutions Outsourcing ____________________________________________ MartiAndersson <hello.martin acomhem.se> Senby: netem-bounces alists.linux-foundation.org 03/26/2008 12:35 PM To netealists.linux-foundation.org cc Subject Re: ratcontrol M Nambiar wrote: > > > Iother words ping responstime = 2*(Size/ABW + Latency) where ABW = > availablbandwidth. > > So for a ping sizof 1400 bytes and 100 ms latency and 64 kbps > bandwidth I expecping responstime to be 2*(175 + 100) = 550ms. > > However I anoquite getting those results at all time when I use > netein conjunction with tbf/htb. > > I used thsamcommands as Tri did except that I used 64kbit rate > instead of 256 kbiin tbf. I goa ping response time of only 337 ms. Aryou doing thnetem command in a route/bridge in between, or is't in thmachinyou're pinging to/from. Sincneteoperate on one path only (175+100=275ms) _______________________________________________ Netemailing list Netealists.linux-foundation.org https://lists.linux-foundation.org/mailman/listinfo/netem ForwardSourceID:NT00018392 =====-----=====-----===== Notice: Thinformation contained in this e-mail messagand/or attachments to imay contain confidential or privileged information. If you ar nothintended recipient, any dissemination, use, review, distribution, printing or copying of th informatiocontained in this e-mail messag and/or attachments to iarstrictly prohibited. If you havreceived this communication in error, pleasnotify us by reply e-mail or telephonand immediately and permanently deletthmessage and any attachments. Thank you -------------- nexpar-------------- AHTML attachmenwas scrubbed... URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20080326/2b5747ae/attachment.ht Fromp.mpatelp agooglemail.com Wed Mar 26 03:49:58 2008 From: mp.mpatelp agooglemail.co(Mittal Patel) Date: Wed, 26 Mar 2008 10:49:58 +0000 Subject: ratcontrol In-Reply-To: <8e5cce0e0803210744u7029cf78m7a1808484bb40fd6@xxxxxxxxxxxxxx> References: <8e5cce0e0803210744u7029cf78m7a1808484bb40fd6@xxxxxxxxxxxxxx> Message-ID: <6fd28abb0803260349r3d57daf1k4526d5e024f4fb6f@xxxxxxxxxxxxxx> Hi Tri, As per # tc qdisc add dev eth0 roohandl1:0 netem delay 100ms you arsetting thdelay to 100ms. so changing the transmission rate of 'tbf' will noaffecmuch. So keep in mind that netem introduce delay ooninterface only i.e. eth0. you caseto netem to introduce delay on reverse path also which mimics original network. Hopthis help. Thank you, Mittal O21/03/2008, Tri Minh Tran <tmt2004 agmail.com> wrote: > hello, > > I gothis examplfrom the netem page to change the transmission rate: > > # tc qdisc add dev eth0 roohandl1:0 netem delay 100ms > # tc qdisc add dev eth0 paren1:1 handl10: tbf rate 256kbit buffer > 1600 limi3000 > # tc -s qdisc ls dev eth0 > qdisc nete1: limi1000 delay 100.0ms > Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) > qdisc tbf 10: rat256Kbiburst 1599b lat 26.6ms > Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) > > I'trying to se'tbf rate' value to different values such as 64kbit, > 128kbit, 256kbiand us'ping' to test it. However I didn't see any > differencin thround trip times of sending packets. > > Could anyontell mif I'm doing anything wrong here? is this the way > to changthtransmission rate? is this the way to test the > transmissioratusing 'ping'? Thanks! > _______________________________________________ > Netemailing list > Netealists.linux-foundation.org > https://lists.linux-foundation.org/mailman/listinfo/netem > Froshemminger alinux-foundation.org Mon Mar 31 08:54:38 2008 From: shemminger alinux-foundation.org (Stephen Hemminger) Date: Mon, 31 Mar 2008 08:54:38 -0700 Subject: ratcontrol In-Reply-To: <8e5cce0e0803210744u7029cf78m7a1808484bb40fd6@xxxxxxxxxxxxxx> References: <8e5cce0e0803210744u7029cf78m7a1808484bb40fd6@xxxxxxxxxxxxxx> Message-ID: <20080331085438.00a1ac20@extreme> OFri, 21 Mar 2008 10:44:59 -0400 "Tri Minh Tran" <tmt2004 agmail.com> wrote: > hello, > > I gothis examplfrom the netem page to change the transmission rate: > > # tc qdisc add dev eth0 roohandl1:0 netem delay 100ms > # tc qdisc add dev eth0 paren1:1 handl10: tbf rate 256kbit buffer > 1600 limi3000 > # tc -s qdisc ls dev eth0 > qdisc nete1: limi1000 delay 100.0ms > Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) > qdisc tbf 10: rat256Kbiburst 1599b lat 26.6ms > Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) > > I'trying to se'tbf rate' value to different values such as 64kbit, > 128kbit, 256kbiand us'ping' to test it. However I didn't see any > differencin thround trip times of sending packets. > > Could anyontell mif I'm doing anything wrong here? is this the way > to changthtransmission rate? is this the way to test the > transmissioratusing 'ping'? Thanks! Do you understand how TokeBuckeworks? It limits the number of packets thacan bsent over a given interval based on rate. TBF does nochangthe actual transmission time of the individual packet. So eveif you seTBF rate to 10 bit/sec, it would just limiyou to on600 bytes/sec or about one full size ethernet packeevery 2 sec, buthe actual time to send that packet would still blinrate so no change in latency due to TBF. Frotmt2004 agmail.com Mon Mar 31 09:09:39 2008 From: tmt2004 agmail.co(Tri Minh Tran) Date: Mon, 31 Mar 2008 12:09:39 -0400 Subject: ratcontrol In-Reply-To: <20080331085438.00a1ac20@extreme> References: <8e5cce0e0803210744u7029cf78m7a1808484bb40fd6@xxxxxxxxxxxxxx> <20080331085438.00a1ac20@extreme> Message-ID: <8e5cce0e0803310909j2750b71fm490a4676bd87a4bc@xxxxxxxxxxxxxx> Hi Stephen, Thank you for your explanation. So could you suggesmhow to change the transmission time of a packeby changing thbandwidth (or transmission rate) using netem? I tried to tesby setting differenvalues 64kbit, 128kbit, 256kbit,... itbf and ping to a remotmachine and did see the differences ithtransmission time for the same packet size. But wheI tried to do thsame but ping to localhost (127.0.0.1) instead of a remotmachine, then I donsee the differences. Could you tell mwhathe problem is? I'd appreciate it. Thanks, Tri OMon, Mar 31, 2008 a11:54 AM, Stephen Hemminger <shemminger alinux-foundation.org> wrote: > OFri, 21 Mar 2008 10:44:59 -0400 > > "Tri Minh Tran" <tmt2004 agmail.com> wrote: > > > > > hello, > > > > I gothis examplfrom the netem page to change the transmission rate: > > > > # tc qdisc add dev eth0 roohandl1:0 netem delay 100ms > > # tc qdisc add dev eth0 paren1:1 handl10: tbf rate 256kbit buffer > > 1600 limi3000 > > # tc -s qdisc ls dev eth0 > > qdisc nete1: limi1000 delay 100.0ms > > Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) > > qdisc tbf 10: rat256Kbiburst 1599b lat 26.6ms > > Sen0 bytes 0 pkts (dropped 0, overlimits 0 ) > > > > I'trying to se'tbf rate' value to different values such as 64kbit, > > 128kbit, 256kbiand us'ping' to test it. However I didn't see any > > differencin thround trip times of sending packets. > > > > Could anyontell mif I'm doing anything wrong here? is this the way > > to changthtransmission rate? is this the way to test the > > transmissioratusing 'ping'? Thanks! > > Do you understand how TokeBuckeworks? It limits the number > of packets thacan bsent over a given interval based on rate. > TBF does nochangthe actual transmission time of the individual > packet. So eveif you seTBF rate to 10 bit/sec, it would just > limiyou to on600 bytes/sec or about one full size ethernet > packeevery 2 sec, buthe actual time to send that packet would > still blinrate so no change in latency due to TBF. > Froshemminger alinux-foundation.org Mon Mar 31 09:48:27 2008 From: shemminger alinux-foundation.org (Stephen Hemminger) Date: Mon, 31 Mar 2008 09:48:27 -0700 Subject: ratcontrol In-Reply-To: <8e5cce0e0803310909j2750b71fm490a4676bd87a4bc@xxxxxxxxxxxxxx> References: <8e5cce0e0803210744u7029cf78m7a1808484bb40fd6@xxxxxxxxxxxxxx> <20080331085438.00a1ac20@extreme> <8e5cce0e0803310909j2750b71fm490a4676bd87a4bc@xxxxxxxxxxxxxx> Message-ID: <20080331094827.72b03d6c@extreme> OMon, 31 Mar 2008 12:09:39 -0400 "Tri Minh Tran" <tmt2004 agmail.com> wrote: > Hi Stephen, > > Thank you for your explanation. > > So could you suggesmhow to change the transmission time of a > packeby changing thbandwidth (or transmission rate) using netem? > > I tried to tesby setting differenvalues 64kbit, 128kbit, > 256kbit,... itbf and ping to a remotmachine and did see the > differences ithtransmission time for the same packet size. But > wheI tried to do thsame but ping to localhost (127.0.0.1) instead > of a remotmachine, then I donsee the differences. Could you tell > mwhathe problem is? I'd appreciate it. > > Thanks, > > Tri I posted aexperimental scheduler thaused fine grain timers, but istill isn'what you want. If you really want to see the effect of packesizon individual packets, you would have to use a physical link (likPPP over serial port) thayou can set hardware rates.
