Hi there,
I made some crude tests Linux-client--Solaris-server and
Solaris-client--Solaris-server. Here's what I got
1) Linux-client--Solaris-server
(cut&paste Linux client)
a) NFS writing
> time dd if=/dev/zero of=tmp/zelo bs=8192 count=10000
> 10000+0 records in
> 10000+0 records out
> 0.050u 4.500s 0:44.49 10.2% 0+0k 0+0io 169pf+0w
That is:
> [~]>bc
> 8192*10000/44.49
> 1841312
1.8 MBytes/sec or...
> (8192*10000/44.49)*8
> 14730496
...14 Mbits/sec for writing
b) NFS reading
> time dd if=tmp/zelo of=/dev/null bs=8192
> 10000+0 records in
> 10000+0 records out
> 0.030u 8.240s 0:19.02 43.4% 0+0k 0+0io 189pf+0w
Meaning:
> 8192*10000/19.02
> 4307045
4.3 Mbytes/sec
> (8192*10000/19.02)*8
> 34456360
or 34 Mbits/sec for reading. _Totally_ different figures than yours..
2) Solaris-client -- Solaris-server
(cut&paste from Solaris client)
a) NFS writing
> time dd if=/dev/zero of=tmp/zelo bs=8192 count=10000
> 10000+0 records in
> 10000+0 records out
> 0.16u 3.39s 0:11.29 31.4%
Meaning
> 8192*10000/11.29
> 7255978
> (8192*10000/11.29)*8
> 58047824
7.2 Mbytes/sec or 58 Mbits/sec for NFS writing
b) NFS reading
> time dd if=tmp/zelo of=/dev/null bs=8192
> 10000+0 records in
> 10000+0 records out
> 0.18u 1.24s 0:11.87 11.9%
Meaning
8192*10000/11.87
6901432
(8192*10000/11.87)*8
55211456
6.9 Mbytes/sec or 55 Mbits/sec NFS reading
Systems: Linux-client: P 200/2.2.5 kernel
Block size: 8kb
Net: 100 Mbit/sec, half-duplex.
Solaris-client: Solaris 2.5.1 on UltraSparc @ 250 Mhz
Solaris-server: Solaris 2.6 / 5.98 on a 2 x UltraSparc @ 250 Mhz
As these tests are quite crude -- i should have tried iozone, bonnie
or smthing. This was just a fast, coffee-break type of benchmark (tm) -- i am
all ears for proposals, but the figures are _so_ blatantly different...
All the best,
Florian
-
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/