... ). This same applies regardless of SACK (no tagging of course). ...Hmm, there's one questionable part ... makes up for the difference you're seeing... > > If you run tcptrace and ... reduced by one and TCP is able to send a new segment out ...

linux-netdev - Ilpo Järvinen - Apr 15 2008 - 04:25

... to the computational overhead of the processing SACK block. All of these systems are multi- ... @vger.kernel.org Subject: Re: A Linux TCP SACK Question Unless you're ... , where the computational overhead of processing SACK blocks is slowing you down, this ...

linux-netdev - Wenji Wu - Apr 4 2008 - 13:49

... that check easier to pass actually: > > Questionable thing is that || in tcp_may_raise_cwnd (might ... Effectively packets_in_flight is reduced by one and TCP is able to > send > ... did give me a good class on Linux ACK/SACK implementation. Thank you. ...

linux-netdev - Wenji Wu - Apr 15 2008 - 14:01

... identical to your setting (one linux sender, one linux receiver, and one ... some loss using a nettem, TCP-SACK showed a bit better ... > At this throughput level, the SACK processing won't take much ... netdev" in the body of a message to majordomo@vger.kernel. ...

linux-netdev - Sangtae Ha - Apr 6 2008 - 18:43

Unless you're sending very fast, where the computational overhead of ... help me out with Linux TCP SACK? Thanks in advance. > > I ... netdev" in > the body of a message to majordomo@vger.kernel. ... netdev" in the body of a message to majordomo@vger.kernel. ...

linux-netdev - John Heffner - Apr 4 2008 - 12:27

... Wenji Wu wrote: > Every system runs Linux 2.6.24. You should have ... the throughput is around 260Mbps Not a surprise, once some reordering is detected, SACK TCP switches away ... unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org ...

linux-netdev - Ilpo Järvinen - Apr 4 2008 - 16:00

... 4 Apr 2008, Wenji Wu wrote: > > > Every system runs Linux 2.6.24. > > You should have reported kernel version right ... sack is off, the throughput is around 260Mbps > > Not a surprise, once some reordering is detected, SACK TCP switches > away > ...

linux-netdev - Wenji Wu - Apr 4 2008 - 16:07

... NIC. As we do a lot of things with SACK, cache misses ... performance. In default setting, I disabled tcp segment offload and did ... you could try also with SACK+basic FRTO (set > >the tcp_frto ... netdev" in > the body of a message to majordomo@vger.kernel. ...

linux-netdev - Sangtae Ha - Apr 5 2008 - 17:17

... are threated differently for NewReno and SACK: NewReno => Sends new data out ( ... not limited to two segments in linux as per RFC, however, RFC doesn' ... arrived while SACK losses its ability to increase outstanding data... ...A quick look ...

linux-netdev - Ilpo Järvinen - Apr 14 2008 - 12:48

... less retransmissions (barely some with > > the SACK in the end). > > > > I suppose the ... saving them from an running TCP flow when a flow ends. > (2) ... tcpdumping the traffic. > > Both experiments with sack off. > > The results is posted to ...

linux-netdev - Ilpo Järvinen - Apr 15 2008 - 04:23

... $ ijjarvin@pointhope:~/linux/debug$ /usr/sbin/tcpdump -n -r sack | grep ... 175233 196953 446558881 3542223511 ijjarvin@pointhope:~/linux/debug$ This is probably far fetched ... unsubscribe netdev" in the body of a message to majordomo@vger.kernel. ...

linux-netdev - Ilpo Järvinen - Apr 8 2008 - 02:36

... of "tcp_ack()", no much processing except sack-tagging the corresponding packets in the ... the window and these ACKs include SACK options, tcp_ack_is_dubious(sk, flag)) would be ... , and include SACK blocks. Not quite sure, just a guess. wenji -- ...

linux-netdev - Wenji Wu - Apr 14 2008 - 18:47

... the same: the throughput > > with SACK on is less than with SACK ... tried with FRTO, FACK. No difference, SACK on is worse than SACK off. wenji -- To unsubscribe from this list: send the line ... netdev" in the body of a message to majordomo@vger.kernel. ...

linux-netdev - Wenji Wu - Apr 4 2008 - 17:40

... less retransmissions (barely some with > the SACK in the end). > > I suppose the ... running and tcpdumping the traffic, run a data transmission test to generate tcp_reordering ... the traffic. Both experiments with sack off. The results is posted to ...

linux-netdev - Wenji Wu - Apr 14 2008 - 18:07

... of the window and these ACKs > include SACK options, tcp_ack_is_dubious(sk, flag)) would be true ... window, and include SACK blocks. > > Not quite sure, just a guess. I had ... going so slowly. You mentioned you're using a web100 kernel. What are the ...

linux-netdev - John Heffner - Apr 14 2008 - 20:48

... , you could try also with SACK+basic FRTO (set >the tcp_frto ... . I tried tcp_frto with 0, 1, 2. SACK On is worse than SACK off. wenji -- To unsubscribe from this list: send the line " ... netdev" in the body of a message to majordomo@vger.kernel. ...

linux-netdev - Wenji Wu - Apr 4 2008 - 18:14

... As we do a lot of things with SACK, cache misses ... performance. > > In default setting, I disabled tcp segment offload and did ... . At this throughput level, the SACK processing won't take much ... netdev" in the body of a message to majordomo@vger.kernel ...

linux-netdev - Wenji Wu - Apr 6 2008 - 16:27

... , you could try also with SACK+basic FRTO (set > >the tcp_frto sysctl ... I tried tcp_frto with 0, 1, 2. > > SACK On is worse than SACK ... we'll have to take a look on tcpdumps. -- i. -- To ... netdev" in the body of a message to majordomo@vger.kernel. ...

linux-netdev - Ilpo Järvinen - Apr 5 2008 - 13:42

... thinking that if the reordered ACKs/SACKs cause confusion in the > sender, ... that you're encountering a bug or thinko in the sack code. This actually brings to mind ... is simpler and more robust than Linux's current reordering threshold calculation. ...

linux-netdev - John Heffner - Apr 4 2008 - 14:07

... result is still the same: the throughput with SACK on is less than with SACK off. wenji -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo ...

linux-netdev - Wenji Wu - Apr 4 2008 - 17:15