Skip to main content

Simple Performance Testing

This section provides an example for testing the performance of Fusion File Share Server to assess its ability to utilize the entire network bandwidth. This test can help identify potential bottlenecks in your environment.

Tools Used:

  • iperf - A popular network performance measurement tool
  • fio - A standard I/O benchmarking tool

Assumptions and Environment:

  • The test is performed on a single Fusion File Share Server, and a single Windows client
  • The server's IP address is 10.210.0.3
  • The server has a share that's mapped to the Z: drive on the client
  • There's no additional network traffic on the server and client, and minimal network traffic on the LAN during the test

First, use iperf to determine the maximum network throughput. Then, test the share's throughput using fio, to evaluate whether it can fully utilize the available network bandwidth.

Network Throughput Test with iperf​

  1. Install iperf on the server and the client:

    • On the server: Use your package manager to install iperf3 (the following example is for Debian or Ubuntu; for other distributions, use the appropriate package manager):

      sudo apt update
      sudo apt install iperf3

    • On the client: Download the latest release of iperf3 for Windows from here, and extract it to C:\iperf3.

  2. Run the test:

    • On the server, in a new terminal, start the iperf3 server: 
      iperf3 -s
    • On the client, open a command prompt, and run the following command to connect to the server (replace 10.210.0.3 with the IP address of your server): 
      C:\> C:\iperf3\iperf3.exe -c 10.210.0.3

The test will run for a few seconds, and its output will resemble:

C:\> iperf3.exe -c 10.210.0.3
Connecting to host 10.210.0.3, port 5201
[  5] local 10.210.0.2 port 49837 connected to 10.210.0.3 port 5201
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec   206 MBytes  1.72 Gbits/sec
[  5]   1.00-2.00   sec   227 MBytes  1.90 Gbits/sec
[  5]   2.00-3.00   sec   230 MBytes  1.93 Gbits/sec
[  5]   3.00-4.01   sec   231 MBytes  1.92 Gbits/sec
[  5]   4.01-5.00   sec   226 MBytes  1.92 Gbits/sec
[  5]   5.00-6.01   sec   238 MBytes  1.98 Gbits/sec
[  5]   6.01-7.00   sec   234 MBytes  1.98 Gbits/sec
[  5]   7.00-8.01   sec   238 MBytes  1.98 Gbits/sec
[  5]   8.01-9.00   sec   233 MBytes  1.97 Gbits/sec
[  5]   9.00-10.01  sec   229 MBytes  1.91 Gbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-10.01  sec  2.24 GBytes  1.92 Gbits/sec                  sender
[  5]   0.00-10.03  sec  2.24 GBytes  1.92 Gbits/sec                  receiver

iperf Done.

From the output, you can determine the maximum network throughput between the server and the client. In this example, the throughput is around 1.92 Gbits/sec, which is about 240 MB/s.

SMB Throughput Test with fio​

To test the throughput of your SMB share, you can use the popular fio tool. It allows you to simulate various I/O patterns and measure the performance of file I/O, making it ideal for evaluating the performance of an SMB server.

The configuration file test.fio you created earlier contains two jobs: one for sequential writes and another for sequential reads. These jobs will write and read 5 GB of data to and from the share (1 GB per job, with 5 jobs each).

  1. Install fio on the client:

    On the client, download the latest release of fio for Windows from here. The installer will extract it to C:\Program Files\fio.

  2. Run the test:

    note

    Starting with Windows 11 24H2 and Windows Server 2025, clients require SMB message signing by default, which might affect client performance and skew the results. To ensure accurate results, disable message signing on the client by running the following PowerShell command:

    Set-SmbClientConfiguration -RequireSecuritySignature $false

    Run the fio tool with the configuration file you've created earlier:

    C:\> Z:
    Z:\> "C:\Program Files\fio\fio.exe" test.fio

The output will resemble:

fio-3.37
Starting 10 threads
sequential-write-direct: Laying out IO file (1 file / 1024MiB)
sequential-write: Laying out IO file (1 file / 1024MiB)
sequential-write: Laying out IO file (1 file / 1024MiB)
sequential-write: Laying out IO file (1 file / 1024MiB)
sequential-write: Laying out IO file (1 file / 1024MiB)
sequential-read: Laying out IO file (1 file / 1024MiB)
sequential-read: Laying out IO file (1 file / 1024MiB)
sequential-read: Laying out IO file (1 file / 1024MiB)
sequential-read: Laying out IO file (1 file / 1024MiB)
sequential-read: Laying out IO file (1 file / 1024MiB)
Jobs: 5 (f=2): [_(5),f(2),R(1),f(1),R(1)][100.0%][r=238MiB/s][r=29 IOPS][eta 00m:00s]
sequential-write-direct: (groupid=0, jobs=10): err= 0: pid=3348: Thu May 23 12:42:54 2024
  read: IOPS=25, BW=205MiB/s (215MB/s)(5120MiB/25026msec)
    slat (usec): min=330, max=8734, avg=678.68, stdev=656.30
    clat (msec): min=92, max=341, avg=189.41, stdev=27.46
     lat (msec): min=92, max=346, avg=190.09, stdev=27.58
    clat percentiles (msec):
     |  1.00th=[  110],  5.00th=[  153], 10.00th=[  163], 20.00th=[  169],
     | 30.00th=[  176], 40.00th=[  182], 50.00th=[  188], 60.00th=[  194],
     | 70.00th=[  203], 80.00th=[  211], 90.00th=[  222], 95.00th=[  232],
     | 99.00th=[  262], 99.50th=[  271], 99.90th=[  342], 99.95th=[  342],
     | 99.99th=[  342]
   bw (  KiB/s): min=97661, max=327045, per=100.00%, avg=213054.79, stdev=9158.66, samples=242
   iops        : min=   11, max=   39, avg=23.82, stdev= 1.21, samples=242
  write: IOPS=27, BW=218MiB/s (229MB/s)(5120MiB/23474msec); 0 zone resets
    slat (msec): min=79, max=353, avg=180.41, stdev=31.09
    clat (nsec): min=634, max=5567.9k, avg=62988.34, stdev=271977.83
     lat (msec): min=79, max=353, avg=180.48, stdev=31.11
    clat percentiles (nsec):
     |  1.00th=[   1012],  5.00th=[   1464], 10.00th=[   1688],
     | 20.00th=[   2064], 30.00th=[   2512], 40.00th=[   3024],
     | 50.00th=[   3856], 60.00th=[  19584], 70.00th=[  29056],
     | 80.00th=[  63744], 90.00th=[ 130560], 95.00th=[ 205824],
     | 99.00th=[ 782336], 99.50th=[1679360], 99.90th=[5537792],
     | 99.95th=[5537792], 99.99th=[5537792]
   bw (  KiB/s): min=144571, max=324207, per=100.00%, avg=226041.45, stdev=8452.16, samples=228
   iops        : min=   14, max=   37, avg=25.31, stdev= 1.11, samples=228
  lat (nsec)   : 750=0.16%, 1000=0.23%
  lat (usec)   : 2=8.67%, 4=16.09%, 10=3.05%, 20=1.80%, 50=8.67%
  lat (usec)   : 100=3.98%, 250=5.39%, 500=1.02%, 750=0.39%, 1000=0.16%
  lat (msec)   : 2=0.31%, 10=0.08%, 100=0.39%, 250=48.52%, 500=1.09%
  cpu          : usr=0.00%, sys=0.00%, ctx=0, majf=0, minf=0
  IO depths    : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=640,640,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=1

Run status group 0 (all jobs):
   READ: bw=205MiB/s (215MB/s), 205MiB/s-205MiB/s (215MB/s-215MB/s), io=5120MiB (5369MB), run=25026-25026msec
  WRITE: bw=218MiB/s (229MB/s), 218MiB/s-218MiB/s (229MB/s-229MB/s), io=5120MiB (5369MB), run=23474-23474msec

From the output, you can determine the read and write throughput of the share.

In this example, the write throughput is approximately 229 MB/s, and the read throughput is around 215 MB/s, indicating a successful result. Ideally, these values should be close to the maximum network throughput determined earlier with iperf, allowing for minor degradation due to network protocol overhead.

Investigating Results​

If the SMB throughput does not closely match the network throughput, consider investigating the following areas: