Skip to main content

Imaging Speed Test for IVS Performance Verification

What

When imaging Windows devices from a FileWave Image Virtualization Server (IVS), it’s important to ensure that data transfer speeds are optimal. Bottlenecks can occur if the IVS and client devices are on different subnets, or if network links between them are not operating at full capacity.

Two tools—iftop and iperf3—can help diagnose and measure network performance for imaging.

Note that if stunnel is enabled then there is some additional bandwidth used for the encrypted tunnel. You can disable stunnel for better performance. 

When/Why

You might use these tests when:

  • Imaging feels slower than expected.

  • Clients are located across multiple switches or VLANs.

  • You want to validate performance before imaging large batches of devices.

  • You need to confirm whether a slowdown is due to the network, the IVS, or client hardware.

How

1. Measuring Real-Time Traffic with iftop

iftop can show live network traffic on the IVS.

Install and run it on the IVS:

Note that in a future version we will include iftop so you'd only need to do the second line to launch it.

sudo apt install iftop
sudo iftop

image.png

The first column will be the server you are running iftop on. The second column will show you the devices connected so if I was imaging 152.32.183.31 then that would be the last entry on the screen and the first line is traffic TO that host and the second is traffic FROM that host as indicated by the arrows. 

Understanding the three traffic columns in iftop:

  • Left column: Average traffic rate over the last 2 seconds.

  • Middle column: Average traffic rate over the last 10 seconds.

  • Right column: Average traffic rate over the last 40 seconds.

You can use this to see per-client speeds during imaging and spot if a client is receiving data more slowly than others.

2. Load Testing with iperf3

iperf3 simulates network traffic to measure throughput between the IVS and a client device.

Install on the IVS:

Note that in a future version we will include iperf3 and it would be running on the IVS already.

The second line below is only needed if you picked "No" to running iperf as a service and will set it to be a service.

sudo apt install iperf3
sudo systemctl enable --now iperf3

Note that iperf3 listens on TCP 5201 so make sure that from your clients you can reach the IVS on that port and you don't have a firewall blocking that port.

On the client side (macOS, Windows, or Debian):

You'll need to first install iperf3. This can be found for Windows here: https://iperf.fr/iperf-download.php and for Debian can be installed with "sudo apt install iperf3" and for macOS can be installed via Homebrew with "brew install iperf3". On macOS many admins will already have Homebrew installed, but it can take a few minutes to install it if not.

iperf3 -c <IVS_IP> -P <parallel_streams> -t <seconds>

  • <IVS_IP>: IP address of the IVS.

  • -P: Number of parallel connections (e.g., 1 for a single client, 20 to simulate 20 machines).

  • -t: Duration in seconds for the test.

Example:

iperf3 -c 192.168.1.50 -P 10 -t 30

Tests 10 parallel streams for 30 seconds.

3. Bandwidth → MB/s Cheat Sheet

(Real-world speeds will be ~10–15% less due to protocol overhead, duplex negotiation, etc.)

Network Speed

Theoretical MB/s

Real-World MB/s (approx.)
1 Gb 125 MB/s 105-115 MB/s
2.5 Gb 312.5 MB/s 265-285 MB/s
5 Gb 625 MB/s 525-565 MB/s
10 Gb 1250 MB/s 1050-1150 MB/s

4. Using Results to Predict Imaging Times

Once you know your average throughput from iperf3, you can estimate imaging time.

Example 1: 1 Gb Network Connection

  • Image: 10 GB

  • Driver pack: 3 GB

  • Total data: 13 GB

  • 1 Gb link ≈ 110 MB/s real-world

Calculation:

13 GB ÷ 110 MB/s ≈ 118 seconds (~2 minutes) per machine.
For 10 machines (unicast), ≈ 20 minutes (plus minor overhead).

Example 2: 10 Gb Network Connection

  • Image: 10 GB

  • Driver pack: 3 GB

  • Total data: 13 GB

  • 10 Gb link ≈ 1100 MB/s real-world

Calculation:

13 GB ÷ 1100 MB/s ≈ 12 seconds per machine.
For 10 machines (unicast), ≈ 2 minutes total.

Expected Imaging Times by Network Speed

(Assumes 13 GB total image + drivers per machine, real-world throughput values, unicast imaging)

Network Speed

Real-World Throughput

1 Machine Time

10 Machines Time

20 Machines Time
1 Gb ~110 MB/s ~2 min ~20 min ~40 min
2.5 Gb ~275 MB/s ~47 sec ~8 min ~16 min
5 Gb ~550 MB/s ~24 sec ~4 min ~8 min
10 Gb ~1100 MB/s ~12 sec ~2 min ~4 min

Additional Notes:

  • Always consider the slowest link: client NIC speed, switch port speed, uplinks between switches, and IVS NIC speed.

  • Avoid hubs—always use switches.

  • Different VLANs and routing hops may reduce performance.

Related Content

Digging Deeper

  • You can combine iftop and iperf3 results: run iperf3 load tests while monitoring iftop to see how the IVS reports real-time speeds.

  • For imaging 50+ devices, consider deploying a local IVS in each subnet to avoid bottlenecks.

  • Remember that hard drives (HDD vs. SSD) in client devices also impact the perceived imaging speed—network speed is only one part of the chain.

  • If speeds are significantly below expected values, investigate:

    • Duplex mismatches on switch ports.

    • Outdated NIC drivers on clients.

    • Oversubscribed uplinks between switches.

     

No comments to display

Back to top