Guides/Test VPS network speed

How to Test VPS Network Speed With iperf3 and Test Files

10gbpsvps ·

Illustration of a bundle of glowing fiber strands fanning out from a single network port

You can measure this network before you pay for anything. Leaseweb runs a public test server in each of the 7 locations, and you can download files from it or run iperf3 against it.

Be clear about what that measures. These tests show the path between your machine and Leaseweb's test server. They do not measure a customer's VPS, and they promise nothing about one. Leaseweb's page about these servers is the source for the files and ports below.

The test servers

LocationTest server and IP
Amsterdamspeedtest.ams1.nl.leaseweb.net
95.211.20.80
Frankfurtspeedtest.fra1.de.leaseweb.net
178.162.226.176
Londonspeedtest.lon1.uk.leaseweb.net
23.106.58.162
Los Angelesspeedtest.lax12.us.leaseweb.net
23.105.41.222
San Franciscospeedtest.sfo12.us.leaseweb.net
23.81.166.99
Washingtonspeedtest.wdc2.us.leaseweb.net
23.105.9.80
Singaporespeedtest.sin1.sg.leaseweb.net
23.108.99.54

Singapore has one catch. The only test server there is in SIN-01, which is a different building from SIN-12, where the VPS runs. In the other seven locations the test server has the same facility code as the VPS.

On 5 November 2026 all four test files answered on all seven hosts, and port 5201 accepted a connection on all seven. A few requests timed out and worked on a second try, so expect an occasional failure.

Download a test file

Each server offers four files at http://HOST/FILE. The names say MB, but the sizes are binary:

FileExact sizeTime at an ideal 10 Gbps
10mb.bin10,485,760 bytesabout 0.008 s
100mb.bin104,857,600 bytesabout 0.08 s
1000mb.bin1,048,576,000 bytesabout 0.84 s
10000mb.bin10,485,760,000 bytesabout 8.4 s

A short download spends much of its time starting up, so it says little about a link of several Gbit/s. At those rates, only the largest file runs long enough to mean much. It also uses over 10 GB of your data, which matters on a metered line.

Use curl rather than a browser. The files are plain HTTP, and browsers may warn about an HTTP download. HTTPS timed out on all seven hosts on 5 November 2026. Writing to /dev/null keeps the file off your disk:

curl -4 -o /dev/null http://speedtest.ams1.nl.leaseweb.net/1000mb.bin

curl's progress meter shows the average speed in bytes per second, with a k or M suffix. This version prints one figure in megabits per second:

curl -4 -s -o /dev/null -w '%{speed_download}\n' \
  http://speedtest.ams1.nl.leaseweb.net/1000mb.bin \
  | awk '{printf "%.0f Mbit/s\n", $1*8/1000000}'

Watch the units. Network speeds are in bits and file transfers usually show bytes, and a byte is 8 bits. So 125 MB per second is 1 Gbit/s.

Run iperf3

iperf3 measures a link without a disk or a web server in the way. You need it installed on your own machine (sudo apt install iperf3 on Ubuntu or Debian).

iperf3 -c speedtest.ams1.nl.leaseweb.net -p 5201 -4

The servers accept iperf3 on ports 5201 to 5210, with one connection per port at a time. If a port is busy, try the next one or wait. -4 forces IPv4, which is the address type each plan includes.

By default your machine sends and the test server receives, so this measures your upload. Add -R to reverse it. The test server then sends and you receive, which is the direction your visitors would see. Add -t 30 to run for 30 seconds in place of the default 10, so the start of the connection counts for less:

iperf3 -c speedtest.ams1.nl.leaseweb.net -p 5201 -4 -t 30 -R

Read the two summary lines at the end, marked sender and receiver. The receiver's bitrate is what arrived. The Retr column counts retransmissions, the packets TCP had to send again. A high count suggests loss somewhere on the path.

Commands for all 7 locations

Run one block at a time. Each curl line downloads about 1 GB, so all seven come to about 7.4 GB, and each iperf3 line sends data for the full 30 seconds.

# Amsterdam, AMS-01
ping -c 20 speedtest.ams1.nl.leaseweb.net
curl -4 -o /dev/null http://speedtest.ams1.nl.leaseweb.net/1000mb.bin
iperf3 -c speedtest.ams1.nl.leaseweb.net -p 5201 -4 -t 30
iperf3 -c speedtest.ams1.nl.leaseweb.net -p 5201 -4 -t 30 -R

# Frankfurt, FRA-01
ping -c 20 speedtest.fra1.de.leaseweb.net
curl -4 -o /dev/null http://speedtest.fra1.de.leaseweb.net/1000mb.bin
iperf3 -c speedtest.fra1.de.leaseweb.net -p 5201 -4 -t 30
iperf3 -c speedtest.fra1.de.leaseweb.net -p 5201 -4 -t 30 -R

# London, LON-01
ping -c 20 speedtest.lon1.uk.leaseweb.net
curl -4 -o /dev/null http://speedtest.lon1.uk.leaseweb.net/1000mb.bin
iperf3 -c speedtest.lon1.uk.leaseweb.net -p 5201 -4 -t 30
iperf3 -c speedtest.lon1.uk.leaseweb.net -p 5201 -4 -t 30 -R

# Los Angeles, LAX-12
ping -c 20 speedtest.lax12.us.leaseweb.net
curl -4 -o /dev/null http://speedtest.lax12.us.leaseweb.net/1000mb.bin
iperf3 -c speedtest.lax12.us.leaseweb.net -p 5201 -4 -t 30
iperf3 -c speedtest.lax12.us.leaseweb.net -p 5201 -4 -t 30 -R

# San Francisco, SFO-12
ping -c 20 speedtest.sfo12.us.leaseweb.net
curl -4 -o /dev/null http://speedtest.sfo12.us.leaseweb.net/1000mb.bin
iperf3 -c speedtest.sfo12.us.leaseweb.net -p 5201 -4 -t 30
iperf3 -c speedtest.sfo12.us.leaseweb.net -p 5201 -4 -t 30 -R

# Washington, WDC-02
ping -c 20 speedtest.wdc2.us.leaseweb.net
curl -4 -o /dev/null http://speedtest.wdc2.us.leaseweb.net/1000mb.bin
iperf3 -c speedtest.wdc2.us.leaseweb.net -p 5201 -4 -t 30
iperf3 -c speedtest.wdc2.us.leaseweb.net -p 5201 -4 -t 30 -R

# Singapore, SIN-12 (the test server is in SIN-01)
ping -c 20 speedtest.sin1.sg.leaseweb.net
curl -4 -o /dev/null http://speedtest.sin1.sg.leaseweb.net/1000mb.bin
iperf3 -c speedtest.sin1.sg.leaseweb.net -p 5201 -4 -t 30
iperf3 -c speedtest.sin1.sg.leaseweb.net -p 5201 -4 -t 30 -R

Why your result may look low

What you seeLikely reasonWhat to do
The result matches your internet planYour own connection is the limit, and it usually isTest from a machine with more bandwidth, such as a server you already rent
iperf3 says the server is busySomeone else is testing on that portTry another port from 5202 to 5210
Results change from run to runLeaseweb says results depend on current load from other tests and downloadsRepeat at different times and compare
Far locations score lower than near onesOne TCP connection slows down as the round trip gets longerSee what a 10 Gbps port means
A request times outThis happens now and thenTry again

Use the looking glass for the path back

The tests above run from you toward Leaseweb. Leaseweb's looking glass runs ping, traceroute and show route from Leaseweb's side. Put in your own public IP address to see the route back to you.

On 5 November 2026 its list of nodes included AMS-01, FRA-01, LON-01, LAX-12, SFO-12 and WDC-02. For Singapore it listed SIN-10, which is not the SIN-12 facility where the VPS runs.

After you order

You can run the same commands from your own server. Remember that data your server sends counts as outbound traffic. An iperf3 upload at 10 Gbps would move 75 GB a minute. That is a small part of 30 TB a month. The traffic guide has the math.

If the numbers suit you, compare the plans. If you are still deciding where to host, the location guide shows how to test from where your users are.