The throughput of local filesystem or an nfs mount can be captured using iozone. The below command starts throughput calculation on /d1 filesystem , with 2 IO threads working on /d1/test/f1 and /d1/test/f2. You can increase the number of threads and filesize.  This will be helpful in analyzing the IO rate on your system of nfs filesystem.

For NFS, you can mount the filesystem with different “rsize” and “wsize” and perform this tests for each rsize/wsize values – and find out the best suited values for your setup.

bash# iozone  -s 100m -t 2 -F /d1/test/f1 /d1/test/f2
        Iozone: Performance Test of File I/O
                Version $Revision: 3.283 $
                Compiled for 32 bit mode.
                Build: linux

        ….

        File size set to 102400 KB
        Command line used: iozone -s 100m -t 2 -F /d1/test/f1 /d1/test/f2
        Output is in Kbytes/sec
        Processor cache size set to 1024 Kbytes.
        Processor cache line size set to 32 bytes.
        File stride size set to 17 * record size.
        Throughput test with 2 processes
        Each process writes a 102400 Kbyte file in 4 Kbyte records

        Children see throughput for  2 initial writers  =   52752.54 KB/sec
        Parent sees throughput for  2 initial writers   =   49633.24 KB/sec
        Min throughput per process                      =   24368.57 KB/sec
        Max throughput per process                      =   28383.98 KB/sec
        Avg throughput per process                      =   26376.27 KB/sec
        Min xfer                                        =   87924.00 KB

        Children see throughput for  2 rewriters        =   58447.32 KB/sec
        Parent sees throughput for  2 rewriters         =   58148.43 KB/sec
        Min throughput per process                      =   28321.71 KB/sec
        Max throughput per process                      =   30125.61 KB/sec
        Avg throughput per process                      =   29223.66 KB/sec
        Min xfer                                        =   96268.00 KB

        Children see throughput for  2 readers          =  597275.21 KB/sec
        Parent sees throughput for  2 readers           =  567971.94 KB/sec
        Min throughput per process                      =   86263.74 KB/sec
        Max throughput per process                      =  511011.47 KB/sec
        Avg throughput per process                      =  298637.61 KB/sec
        Min xfer                                        =   18244.00 KB

        …

        Children see throughput for 2 reverse readers   =  551658.72 KB/sec
        Parent sees throughput for 2 reverse readers    =  549420.03 KB/sec
        Min throughput per process                      =    9038.66 KB/sec
        Max throughput per process                      =  542620.06 KB/sec
        Avg throughput per process                      =  275829.36 KB/sec
        Min xfer                                        =    1708.00 KB

        Children see throughput for 2 stride readers    =  377521.64 KB/sec
        Parent sees throughput for 2 stride readers     =  368290.16 KB/sec
        Min throughput per process                      =   59013.74 KB/sec
        Max throughput per process                      =  318507.91 KB/sec
        Avg throughput per process                      =  188760.82 KB/sec
        Min xfer                                        =   19472.00 KB

        Children see throughput for 2 random readers    =  352898.12 KB/sec
        Parent sees throughput for 2 random readers     =  351727.80 KB/sec
        Min throughput per process                      =   27341.34 KB/sec
        Max throughput per process                      =  325556.78 KB/sec
        Avg throughput per process                      =  176449.06 KB/sec
        Min xfer                                        =    8612.00 KB

        Children see throughput for 2 mixed workload    =  177942.61 KB/sec
        Parent sees throughput for 2 mixed workload     =   89133.63 KB/sec
        Min throughput per process                      =   64354.38 KB/sec
        Max throughput per process                      =  113588.23 KB/sec
        Avg throughput per process                      =   88971.30 KB/sec
        Min xfer                                        =   58012.00 KB

          …

        Children see throughput for 2 pwrite writers    =   55950.09 KB/sec
        Parent sees throughput for 2 pwrite writers     =   55377.80 KB/sec
        Min throughput per process                      =   27811.84 KB/sec
        Max throughput per process                      =   28138.25 KB/sec
        Avg throughput per process                      =   27975.04 KB/sec
        Min xfer                                        =  101212.00 KB

        Children see throughput for 2 pread readers     = 1165896.81 KB/sec
        Parent sees throughput for 2 pread readers      = 1128007.09 KB/sec
        Min throughput per process                      =  577944.88 KB/sec
        Max throughput per process                      =  587951.94 KB/sec
        Avg throughput per process                      =  582948.41 KB/sec
        Min xfer                                        =   94876.00 KB

iozone test complete.

0 Shares:
Leave a Reply

Your email address will not be published. Required fields are marked *

You May Also Like
Read More

MobileRack

Rackable systems has released mobile rack – a mobile datacenter. This MobiRack is a compact , light-weight plastic…
Read More

Netapp VIFs loadbalancing

NetApp loadbalancing algorithm for determining ethernet load balancing (vifs) differs from Cisco, so it’s normal to see unbalanced…