To make everyone happy: Netperf!
First Test Solaris Netra T1 <-> Netra T1 100MGQF FD Flip Cable. Solaris
2.7
# ./netperf -H 172.16.10.1 -l 60
TCP STREAM TEST to 172.16.10.1
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec
65535 32768 32768 60.00 94.16
#
Second Test: Solaris 2.6 Ultra2-Dual Processor 768Mram Talking to a
netserver on itself. (This is also the source machine of the ratbert dump
from before)
> ./netperf -H 172.16.10.2 -l 60
TCP STREAM TEST to 172.16.10.2
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec
65536 65536 65536 60.04 105.92
Third Test - Above ratbert box talking to Linux 2.2.14 box With
EtherExpressPro 100 As in the original test.. (Shared Media, Switched)
> ./netperf -H newspeer2.tds.net -l 60
TCP STREAM TEST to newspeer2.tds.net
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec
65535 65536 65536 60.00 39.20
Third Test - Above ratbert box talking to Linux 2.2.14 box With
EtherExpressPro 100 As in the original test.. (Private Media, Flipped)
> ./netperf -H 172.16.10.1 -l 60
TCP STREAM TEST to 172.16.10.1
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec
65535 65536 65536 60.00 43.63
Is everyone happier now :) Any other tests or option version to try with
netperf. Would you like any other tcpdumps, I am also willing to give some
of the kernel people like alan, jes, or james access to the linux box to
check it out. via ssh.
On Thu, 4 May 2000, Alan Cox wrote:
> > shane> Not Much Difference :(
> >
> > You're using WHAT as a benchmark?
> >
> > I am sorry but ftp is like the stupidiest thing to use for any sort of
>
> Its actually not too silly in this case. Trying a transfer into /dev/null
> may be a good idea for testing the 'is this disk or net' thats the bottleneck
> and doing a ttcp test to compare is wise.
>
> THe ncftp client should be quite sane
>
> Alan
>
>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun May 07 2000 - 21:00:15 EST