Date: Sat, 9 Nov 2013 14:24:52 -0800 From: Adrian Chadd <adrian@freebsd.org> To: Oleg Moskalenko <mom040267@gmail.com> Cc: FreeBSD Net <freebsd-net@freebsd.org> Subject: Re: Question about network stack advancements to be on the same level as Linux kernel 3.9+ Message-ID: <CAJ-Vmono%2BHtaKV%2BvsaKncDByuvV1jUrsgAFq7NYvOG=JPbNg3Q@mail.gmail.com> In-Reply-To: <CALDtMrLAeWekT1Zp%2BEt8r%2BWzF2V2gruka_M%2B%2Bk2yxq=AvXv8Gw@mail.gmail.com> References: <CALDtMr%2B%2B7uYiJWyhEdYOR5vxhGSJPKFXXc8L%2BKRaVctUnZaiKA@mail.gmail.com> <CAJ-VmokOp8Jo8giwnmis7VQwB6KJLAk6BwPeqw0ppZjkz7Wtsg@mail.gmail.com> <CALDtMrLAeWekT1Zp%2BEt8r%2BWzF2V2gruka_M%2B%2Bk2yxq=AvXv8Gw@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 9 November 2013 13:58, Oleg Moskalenko <mom040267@gmail.com> wrote: > Those kind of advancements make much more sense for UDP than for TCP. > > Of course TCP would benefit from them, too, but they are really critical for > UDP. > > TCP stack is already relatively advanced, and the improvements will help in > only some really extreme high-load use cases - when the TCP listener is the > bottleneck. On the other hand, UDP would benefit from the improvements even > in usual ordinary use cases. If I may suggest the priority, it would make > much more sense to start improving the stack with the UDP. I don't mind which one gets done, as long as one of them does. :-) I haven't yet written much UDP testing code. I'll be doing it soon. I may even take a stab at the UDP side of things. But I'm still knee deep in mbuf stuff at work (and wifi stuff at home) so I can't make any guarantees. -adrian
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-Vmono%2BHtaKV%2BvsaKncDByuvV1jUrsgAFq7NYvOG=JPbNg3Q>