Test Performance

Test Performance

This section will test the performance of STG326 - SAZ.

There are many disk performance tools out there. We have pre-installed DiskSpd and fio on the Windows Instance 0 instance. If you are familiar with another tool and would prefer using it to test performance, feel free to download and use it.

$client = new-object System.Net.WebClient


$client.DownloadFile("https://lab.com/microsoft/diskspd/releases/download/v2.0.21a/DiskSpd.zip","C:\Tools\DiskSpd-2.0.21a\DiskSpd-2.0.21a.zip")


Expand-Archive -LiteralPath C:\Tools\DiskSpd-2.0.21a\DiskSpd-2.0.21a.zip -DestinationPath C:\Tools\DiskSpd-2.0.21a

Note

It will take approximately 30 minutes to complete this section.

Important

Read through all steps below and watch the quick video before continuing.

test-performance.gif

DiskSpd Read tests

  1. Access the remote desktop session to Windows Instance 0.

  2. Click Start >> Windows PowerShell.

    Important

    This section assumes that STG326 - SAZ is mapped as the Z:/ drive. If your Windows Instance 0 does not have a mapped Z:/ drive, map STG326 - SAZ as the Z:/ drive (see the previous section for step-by-step instructions).

  3. Run the script below in the PowerShell seccion to create a 100 GB sparse file.

$random = $(Get-Random)
fsutil file createnew Z:\${env:computername}-$random.dat 100000000000
  1. Runthe DiskSpeed script below to test read performance of the mapped Z: drive
C:\Tools\DiskSpd-2.0.21a\amd64\DiskSpd.exe -d120 -w0 -r -t1 -o32 -b64K -Su -L Z:\${env:computername}-$random.dat

Amazon Fsx for Windows Server

While the script is running, open Task Manager and monitor network performance (e.g. Start >> Task Manager >> More details>> Performance (tab) >> Ethernet).

  • What is the baseline throughput of the file system? - This was configured when the file system was created and is available from the Amazon FSx console.

    • What was the peak read throughput you achieved?

    • What was the P99 (99th %-tile) of your test?

    • What was the Total Read IO MiB/s?

    • What was the I/O per second?

    • What was the AvgLat?

    • Why is your current throughput greater than the baseline throughput?

Amazon Fsx for Windows Server

Amazon Fsx for Windows Server

  1. Experiment with different DiskSpd parameter settings. Use the table below as a guide. Test with different block sizes (-b), number of outstanding I/O requests (-o), number of threads per file (-t), and enable/disable software caching (-Su).
    Parameter Description
    -b<size>[K|M|G] Block size in bytes or KiB, MiB, or GiB (default = 64K).
    -o<count> Number of outstanding I/O requests per-target per-thread. (1 = synchronous I/O, unless more than one thread is specified with by using -F.) (default = 2)
    -r<size>[K|M|G] Random I/O aligned to the specified number of bytes or KiB, MiB, GiB, or blocks. Overrides -s (default stride = block size).
    -s<size>[K|M|G] Sequential stride size, offset between subsequent I/O operations in bytes or KiB, MiB, GiB, or blocks. Ignored if -r is specified (default access = sequential, default stride = block size).
    -t<count> Number of threads per target. Conflicts with -F, which specifies the total number of threads.
    -Su Disable software caching.

    • What different parameters did you test?

    • How did the different parameter options alter the results?

DiskSpd Write tests

  1. From the remote desktop session to Windows Instance 0, open a PowerShell window.

    Important

    This section assumes that STG326 - SAZ is mapped as the Z:/ drive. If your Windows Instance 0 does not have a mapped Z:/ drive, map STG326 - SAZ as the Z:/ drive (see the previous section for step-by-step instructions).

  2. Run the DiskSpeed script below to test write performance of the mapped Z: drive

$random = $(Get-Random)
C:\Tools\DiskSpd-2.0.21a\amd64\DiskSpd.exe -d120 -c2G -s64K -w100 -t1 -o32 -b64K -Sh -L Z:\${env:computername}-$random.dat

Amazon Fsx for Windows Server

While the script is running, open Task Manager and monitor network performance (e.g. Start >> Task Manager >> Performance (tab) >> Ethernet).

  • What is the baseline throughput of the file system? - This was configured when the file system was created and is available from the Amazon FSx console.

    • What was the peak write throughput you achieved?

    • What was the P99 (99th %-tile) of your test?

    • What was the Total Write IO MiB/s?

    • What was the I/O per second?

    • What was the AvgLat?

    • Why is your current throughput greater than the baseline throughput?

  1. Experiment with different DiskSpd parameter settings. Use the table below as a guide. Test with different file sizes (-c), different block sizes (-b), number of outstanding I/O requests (-o), number of threads per file (-t), random I/O (-r) or sequential I/O (-s), and enable/disable software & hardware write caching (-Sh).
    Parameter Description
    -d<seconds> Duration in seconds.
    -b<size>[K|M|G] Block size in bytes or KiB, MiB, or GiB (default = 64K).
    -c<size>[K|M|G] Create files of the specified size. Size can be stated in bytes or KiBs, MiBs, GiBs, or blocks.
    -r<size>[K|M|G] Random I/O aligned to the specified number of bytes or KiB, MiB, GiB, or blocks. Overrides -s.
    -s<size>[K|M|G] Sequential stride size, offset between subsequent I/O operations in bytes or KiB, MiB, GiB, or blocks. Ignored if -r is specified (default access = sequential, default stride = block size).
    -o<count> Number of outstanding I/O requests per-target per-thread. (1 = synchronous I/O, unless more than one thread is specified with by using -F.) (default = 2)
    -t<count> Number of threads per target. Conflicts with -F, which specifies the total number of threads.
    -Sh Disables both software caching and hardware write caching.

    • What different parameters did you test?

    • How did the different parameter options alter the results?

fio read tests

  1. From the remote desktop session to Windows Instance 0, open a PowerShell window.

    Important

    This section assumes that STG326 - SAZ is mapped as the Z:/ drive. If your Windows Instance 0 does not have a mapped Z:/ drive, map STG326 - SAZ as the Z:/ drive (see the previous section for step-by-step instructions).

  2. Run the fio script below to test read performance of the mapped Z: drive

$random = $(Get-Random)
C:\Tools\fio-3.16-x64\fio --randrepeat=1 --direct=1 --name="Z:\${env:computername}-$random.dat" --numjobs=1 --bs=64K --iodepth=32 --size=1024M --readwrite=read --rwmixread=100 --thread --time_based --runtime=120

While the script is running, open Task Manager and monitor network performance (e.g. Start >> Task Manager >> Performance (tab) >> Ethernet).

  • What was the peak read throughput?
  • What was the average read throughput?
  • What was the average IOPS?
  • How many GB did you read in 120 seconds?
  1. Experiment with different fio parameter settings. Use the table below as a guide. Test with direct I/O enabled or disabled (–direct), different block sizes (–bs), number of outstanding I/O requests (–iodepth), number of jobs (–numjobs), random read, random write, sequential read, sequential write (–readwrite), and mixture of reads and writes (–rwmixread).
Parameter Description
--direct=[0|1] Use buffered (0) or non-buffered (1) I/O.
--bs=<size>[K|M|G] Block size in bytes or KiB, MiB, or GiB (default = 64K).
--numjobs=<count> Number of clones (processes/threads performing the same workload) of this job. Default: 1.
--readwrite=[read|write|randread|randwrite] Type of I/O pattern (read = sequential read; write = sequential write; randread = random read; randwrite = random write).
--iodepth=<count> Number of I/O units to keep in flight against the file.
--rwmixread=<percent> Percentage of a mixed workload that should be reads. The outstanding percentage will be writes.

  • What different parameters did you test?

  • How did the different parameter options alter the results?

fio write tests

  1. From the remote desktop session to Windows Instance 0, open a PowerShell window.

    Important

    This section assumes that STG326 - SAZ is mapped as the Z:/ drive. If your Windows Instance 0 does not have a mapped Z:/ drive, map STG326 - SAZ as the Z:/ drive (see the previous section for step-by-step instructions).

  2. Run the fio script below to test write performance of the mapped Z: drive

$random = $(Get-Random)
C:\Tools\fio-3.16-x64\fio --randrepeat=1 --direct=1 --name="Z:\${env:computername}-$random.dat" --numjobs=1 --bs=64K --iodepth=32 --size=1024M --readwrite=write --rwmixwrite=100 --thread --time_based --runtime=120

While the script is running, open Task Manager and monitor network performance (e.g. Start >> Task Manager >> Performance (tab) >> Ethernet).

  • What was the peak write throughput?
  • What was the average write throughput?
  • What was the average IOPS?
  • How many GB did you write in 120 seconds?
  1. Experiment with different fio parameter settings. Use the table below as a guide. Test with direct I/O enabled or disabled (–direct), different block sizes (–bs), number of outstanding I/O requests (–iodepth), number of jobs (–numjobs), random read, random write, sequential read, sequential write (–readwrite), and mixture of reads and writes (–rwmixwrite).
Parameter Description
--direct=[0|1] Use buffered (0) or non-buffered (1) I/O.
--bs=<size>[K|M|G] Block size in bytes or KiB, MiB, or GiB (default = 64K).
--numjobs=<count> Number of clones (processes/threads performing the same workload) of this job. Default: 1.
--readwrite=[read|write|randread|randwrite] Type of I/O pattern (read = sequential read; write = sequential write; randread = random read; randwrite = random write).
--iodepth=<count> Number of I/O units to keep in flight against the file.
--rwmixwrite=<percent> Percentage of a mixed workload that should be writes. The outstanding percentage will be reads

  • What different parameters did you test?

  • How did the different parameter options alter the results?

    1. Close all the PowerShell windows. Run exit.

    2. Close the File Explorer window.

    3. Close the Task Manager window.