Hi you are experiencing performance drop with smaller packets because you have more packets per second, thus more memcpy() call’s and more per-packet overhead, let’s say it is expected for this application, you should try to redesign it (using multiple threads for instance).
Alfredo > On 11 Jun 2015, at 17:24, [email protected] wrote: > > ok.I cross connect to 10GE port and use "pfsend -i dna0 -g 8 -l 64" to send > packets on dna0, while use "silidump -i dna1 -g 15 -w /mnt/disk1/" to capture > packets and save to disk. silidump written by my self reference from pfdump > demo code. pfsend send packets at speed of nearly 13Mpps so i am sure of the > pf_ring.ko and ixgbe.ko(DNA) was loaded corectly.silidump create on thread > dedicated capture pakcet by calling pfring_recv(pd, &buff, 0, &hdr, 1) within > a dead loop, and copy buff to another 2GB memaligned buffer while receive one > packet. in this manner pfring_stats shows packet drop. > key code like this: > for(;;){ > if(pfring_recv(&pd, &buff, 0, &hdr, 1)>0){ > memcpy(buff, hdr.len, alignedbuffer); > ++pkt_cnt; > } > } > without memcpy pfring_stats shows no packet drop. > my server os is ubuntu 12.04 64bit version running on 2 xeon E2470 @2.3GHz > each has 8 core. > give any clue if you can. thanks and best regards :-) >> Hi, please explain in detail the tests you have made with complete command >> line options Luca > On 05 Jun 2015, at 02:02, tiankun zhang >> <zhangtiankun1988 at gmail.com >> <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>> wrote: > > >> appreciate your reply luca. one more question, i want to do packet > capture >> at line rate using PF_RING+DNA, i tested my program with 512 byte packet at >> 10Gbps , there's no packet drop. BUT with 256, 128,64 byte packet drop >> increased dramatically. actually, i use memcpy to copy packet to user space >> buffer. may this be the problem at high pps ? any idea for fixing it? >> looking forward reply thx 😊 > > tiankun zhang <zhangtiankun1988 at gmail.com >> <http://listgateway.unipi.it/mailman/listinfo/ntop-misc> >> <mailto:zhangtiankun1988 at gmail.com >> <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>>>于2015年6月4日星期四写道： > >> > _______________________________________________ > Ntop-misc mailing list > >> Ntop-misc at listgateway.unipi.it >> <http://listgateway.unipi.it/mailman/listinfo/ntop-misc> > >> http://listgateway.unipi.it/mailman/listinfo/ntop-misc >> <http://listgateway.unipi.it/mailman/listinfo/ntop-misc> -------------- next >> part -------------- An HTML attachment was scrubbed... URL: >> <http://listgateway.unipi.it/mailman/private/ntop-misc/attachments/20150608/640865e0/attachment.htm >> >> <http://listgateway.unipi.it/mailman/private/ntop-misc/attachments/20150608/640865e0/attachment.htm>> > _______________________________________________ > Ntop-misc mailing list > [email protected] > http://listgateway.unipi.it/mailman/listinfo/ntop-misc
_______________________________________________ Ntop-misc mailing list [email protected] http://listgateway.unipi.it/mailman/listinfo/ntop-misc
