Date: Fri, 25 Mar 2011 08:47:44 -0700 From: Rumen Telbizov <telbizov@gmail.com> To: Freddie Cash <fjwcash@gmail.com> Cc: Denny Schierz <linuxmail@4lin.net>, freebsd-stable <freebsd-stable@freebsd.org> Subject: Re: SAS HBA LSI 9200-8e supported under 8.2? Message-ID: <AANLkTincriB%2B4%2B4H4hqhwWj_u3JABm6DpBzKSCbGboh9@mail.gmail.com> In-Reply-To: <AANLkTikxOiq_H9M_EU6dLAQ3jt=8ynaD_dyCs2t5-7rL@mail.gmail.com> References: <1300618959.2191.5.camel@netti> <AANLkTi=8%2BrM9CJUd2ACB8U1QA9LzBwcwBLg0LjAwOTpH@mail.gmail.com> <1300887484.5930.72.camel@pcdenny> <AANLkTinGw0Lbnhu%2Bbo01RiRgPN1TpBouG1XdxM4T9ZOz@mail.gmail.com> <8CDBE9B8-4834-4F2C-84B9-2DA371D5B2C3@4lin.net> <AANLkTik8NmvRdH_SfXzcvTxVvhy8n77Q80hbaXniLFBQ@mail.gmail.com> <1301056115.21588.103.camel@pcdenny> <AANLkTikxOiq_H9M_EU6dLAQ3jt=8ynaD_dyCs2t5-7rL@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Good job Denny, I am happy to hear you made it work. On Fri, Mar 25, 2011 at 7:38 AM, Freddie Cash <fjwcash@gmail.com> wrote: > On Fri, Mar 25, 2011 at 5:28 AM, Denny Schierz <linuxmail@4lin.net> wrote: > > So, how complicated is load balancing with two Gb Network cards? :-) Ok, > > that should be a new thread;-) > > Simple as pie. Read through lagg(4) to see how it's done from the > command-line using ifconfig(8). > > ifconfig_lagg0="laggproto round-robin laggport em0 laggport em1 inet > 192.168.0.1/24" > That's actually an interesting problem (although it really belongs to a different thread). Freddie, have you tried this with an HP Procurve (say 2910al) switch. I did and a dual gigabit connection between two machines made the switch cpu utilization to jump from around 1% to 30%. I guess it might be HP specific problem due to the enormous mac address bounce between the two ports. I am curious to know if anybody else has experienced similar problems? It's a really important problem to solve since having zpool's that can achieve a gigabyte a second transfers makes little sense when you are limited by a gigabit (120MB/s) network connection to the rest of the world. Alternative to round-robin is only maybe LACP, but it doesn't really cut it when it comes to transfers between two nodes only due to the hashing that LACP uses. Or maybe 10GbE anyone? Cheers, -- Rumen Telbizov http://telbizov.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?AANLkTincriB%2B4%2B4H4hqhwWj_u3JABm6DpBzKSCbGboh9>