Jaime nice to hear from you again after so long time. On May 5, 2009, at 8:39 PM, Jaime Nebrera wrote:
> Hi Luca, > > As always, get all my cheers for such a great job. We will purchase > it > ASAP. > > Some questions / suggestions: > > 1) When you purchase the 10GE version, is the 1GE included? No as they are two different drivers > > 2) It would be quite clarifying if in the web page you have created > for the software you include a drawing not only of current situation > but > also a couple of new ones detailing the new "proposed" options (either > just multi thread at the driver level or at the application level in > combination with pfring) I think this will clarify quite a lot. I'll do this tonight > > 3) We hope to make some benchmarking on our own hardware. Currently > we > have a supported 1GE box we could provide for testing (including a > Quad > Core Xeon) You should be able to capture packets at wire rate any packet size > > 4) Last but not least, are you planning to support other "advanced" > 10GE drivers, say Chelsio 10GE? The point is, we are going to be > able to > put our hands pretty soon in a lot of equipment using this card (IBM > BladeCenter H with HS22 blades and Cheslio 2x10GE cards). We could use > this hardware to be able to stress and enhance all this wonderful > software. I understand it might not be possible as in essence means a > different stuff and things learned cannot be applied to the Intel > side, > but I think it should). > > Sadly there are no Intel based cards for IBM BladeCenter, so we cant > provide them :) > > I understand Chelsio card should include quite a lot of Intel > advanced > capabilities "in hardware". I can provide you a contact within Chelsio > to discuss this topic if you want. I don't know Cheslio cards. All I say is that writing patches for kernel drivers is an activity that takes a lot of time as it requires low-level card knowledge. Said that I don't think that this is gonna be impossible but definitively time consuming. Instead adding native PF_RING support, preserving RX multiqueue (but not implementing threaded packet capture) should be pretty simple. > > I will discuss on this topic more deeply privately. > Thanks Luca > Congratulations > > El mar, 05-05-2009 a las 00:37 +0200, Luca Deri escribió: >> Dear all >> after a couple of years of research, I have the pleasure to announce >> TNAPI a new driver approach (currently available for modern Intel 1 >> and 10 Gbit cards) that by leveraging PF_RING multi RX-queue support, >> it allows packet capture speed to be greatly improved. Wire-rate >> packet packet capture is now feasible using commodity hardware. >> >> You can find all TNAPI information at http://www.ntop.org/TNAPI.html. >> Please note that contrary to PF_RING, TNAPI is driver-dependent hence >> make sure your ethernet NIC is compatible with TNAPI before thinking >> to use it. >> >> >> Enjoy, Luca >> _______________________________________________ >> Ntop-misc mailing list >> [email protected] >> http://listgateway.unipi.it/mailman/listinfo/ntop-misc > -- > Jaime Nebrera - [email protected] > Consultor TI - ENEO Tecnologia SL > Pol. PISA - C/ Manufactura 6, P1, 3B > Mairena del Aljarafe - 41927 - Sevilla > Telf.- 955 60 11 60 / 619 04 55 18 > > _______________________________________________ > Ntop-misc mailing list > [email protected] > http://listgateway.unipi.it/mailman/listinfo/ntop-misc _______________________________________________ Ntop mailing list [email protected] http://listgateway.unipi.it/mailman/listinfo/ntop
