gigabit xfer speeds under linux
aurbain at staff.theuseful.com
Tue May 18 14:09:12 CDT 2004
Rogers, Jamie wrote:
>I curious about your kernel/TCP setting to achieve your throughput. Can you
>list your relevant parameters and their settings?
>From: Sean Dilda [mailto:agrajag at dragaera.net]
>Sent: Tuesday, May 18, 2004 2:35 PM
>To: Aaron Urbain
>Cc: linux-poweredge at dell.com
>Subject: Re: gigabit xfer speeds under linux
>On Tue, 2004-05-18 at 14:00, Aaron Urbain wrote:
>>As an aside, does anyone know whether the Broadcom cards in a 1750 are
>>I have another server (not dell) with the same cards (5704) and
>>data between them using the Cisco (everything at 1000/FD) I'm only seeing
>>30MB/sec (~240mbit) which is a bit pants really for GigE...
>>(Second server is a Dual Opteron 248)
>That is pretty slow for a gigabit. However, its a believable transfer
>speed for a harddrive.
>If have a cluster of 1750's, all connected with some powerconnect
>5224's. Using ttcp transfering 16gigabyte chunks, I've gotten around
>112MB/sec, both ways.
>Linux-PowerEdge mailing list
>Linux-PowerEdge at dell.com
>Please read the FAQ at http://lists.us.dell.com/faq or search the list
>archives at http://lists.us.dell.com/htdig/
sysctl -w net.core.rmem_max=8388608
sysctl -w net.core.wmem_max=8388608
sysctl -w net.core.rmem_default=65536
sysctl -w net.core.wmem_default=65536
sysctl -w net.ipv4.tcp_rmem='4096 87380 8388608'
sysctl -w net.ipv4.tcp_wmem='4096 65536 8388608'
sysctl -w net.ipv4.tcp_mem='8388608 8388608 8388608'
sysctl -w net.ipv4.route.flush=1
running the above gives me a few more meg/sec when I netcat data.
am reading and writing u320 raid0 of 6 disks, adaptec 2200s.
ttcp gives me 90MB/s
More information about the Linux-PowerEdge