[TriLUG] NFS write performance problem

[John Franklin]

On Fri, Dec 27, 2002 at 02:08:11PM -0500, Scott Stancil wrote:
> I am having NFS write performance issues with a backend dedicated NFS
> server (RedHat 8.0, PIII 550, 128MB ram, 100Mbps network).  The client (RH
> 6.1, Kernel 2.2.19SMP with nfs-utils-0.1.9.1-1, dual PIII 550's, 1GB ram,
> 100Mbps network) is mounting the home directory off of the server. 
> Integrated Intel EtherExpress 100 ethernet devices on both.
> 
> Testing Methods:
> I am using the following to test the network transfers from the client to
> the server.
>  Write:
>  time dd if=/dev/zero of=/mnt/home/testfile bs=x count=y
> 
>  Read:
>  time dd if=/mnt/home/testfile of=/dev/null bs=x
> 
> 
> Results:
> Unmounts between changes to clear caching. Reading and writing a 262MB file.
> 
> X=8 Y=32768, Write=2:48, Read=0:10, 1.55 MB/second, 24.25MB/second read
> 
> X=16 Y=16384, Write=2:48, Read=0:25, 1.55 MB/second write, 10.5 MB/second
> 
> read X=32 Y=8291, Write=2:48, Read=0:15, 1.55MB/second write, 17.5
> MB/second read

Is that really "8" (e.g. dd if=/dev/zero of=testfile bs=8 count=32768)
and not "8k"?

What kind of times do you get with x=8k x=16k x=64k?

Here's my version of the test:

Client: Dual P2/266 NetBSD 1.6K 447M 3Com 3c905-TX
Server: Single P1/100 NetBSD 1.6 (release) 96M Lite-On 82C169 Ethernet

[4]% time dd if=/dev/zero of=testfile bs=8k count=32768
32768+0 records in
32768+0 records out
268435456 bytes transferred in 51.717 secs (5190468 bytes/sec) <- 5 M/s
0.382u 13.307s 0:51.75 26.4%    0+0k 0+0io 1pf+0w

Your systems should vastly surpass mine, but:

[8]% time dd if=/dev/zero of=testfile bs=8 count=32768
32768+0 records in
32768+0 records out
262144 bytes transferred in 2.800 secs (93622 bytes/sec)  <- 93 K/s
0.134u 2.547s 0:02.89 92.3%     0+0k 0+0io 0pf+0w
[9]% time dd if=/dev/zero of=testfile bs=16 count=16384
16384+0 records in
16384+0 records out
262144 bytes transferred in 1.537 secs (170555 bytes/sec) <- 170 K/s
0.106u 1.234s 0:01.55 85.8%     0+0k 0+0io 0pf+0w

( "<- " annotations mine)

1.5M/s on your system I could believe with an 8 byte blocksize, but not
an 8k blocksize.

jf
-- 
John Franklin
franklin at elfie.org
ICBM: 35°43'56"N 78°53'27"W