Are all Ookla testservers equal?

TL;DR

With the same original Speedtest by Ookla, we tested our internet speed with the 10 suggested local servers and measured different speeds.

The slowest test server gives us 100.41 Mbps, the fastest test server 101.95 Mbps.

We cannot properly explain these results and are puzzled.

 

Contents

  1. Introduction
  2. Method of measurement
  3. Test servers to test
  4. The measurements
  5. Conclusions
 

Introduction

When you use an Ookla Speedtest you might select a testserver. The study A Comparative Analysis of Ookla Speedtest and Measurement Labs Network Diagnostic Test (NDT7) (PDF) states, among other things, the following: (...) For Ookla, the choice of test server can significantly affect the reported speed. Tests using certain Ookla servers systematically report speeds 10% lower than other servers (...).

We haven't noticed such big differences in the test server used in an Ookla speed test. All the more reason to find out how consistent the Ookla test servers are. Hence, we wonder if the test results depends on the testserver used.

 

Method of measurement

For this test we use the original Ookla Speedtest, and run 5 times a speed test per local server. Because we have a 100 Mbps plan we expect to measure 100.8 Mbps.

 

Test servers to test

We test all local servers as suggested by the Ookla Speedtest.

 

The measurements

  1. Rotterdam - Sixteen-Hosting 100.52, 100.50, 100.53, 100.44 and 100.52 Mbps. Average: 100.50 Mbps, standard deviation: 0.04 Mbps
  2. Amsterdam - Odido 100.47, 100.49, 100.48, 100.10 and 100.51 Mbps. Average: 100.41 Mbps, standard deviation: 0.17 Mbps
  3. Amsterdam - 31173 Services AB 101.91, 101.90, 101.92, 101.92 and 101.91 Mbps. Average: 101.91 Mbps, standard deviation: 0.01 Mbps
  4. Zoetermeer - ColoCenter bv 101.84, 101.89, 101.98, 101.93 and 101.97 Mbps. Average: 101.92 Mbps, standard deviation: 0.06 Mbps
  5. Naaldwijk - Host Sailor Ltd 100.40, 100.54, 100.53, 100.54 and 100.45 Mbps. Average: 100.49 Mbps, standard deviation: 0.06 Mbps
  6. Naaldwijk - WorldStream B.V. 100.51, 100.49, 100.44, 100.51 and 100.49 Mbps. Average: 100.49 Mbps, standard deviation: 0.03 Mbps
  7. Amstelveen - KPN B.V. 100.60, 100.55, 100.53, 100.49 and 100.5 Mbps. Average: 100.53 Mbps, standard deviation: 0.04 Mbps
  8. Amsterdam - Labixe Ltd 100.51, 100.49, 100.55, 100.48 and 100.59 Mbps. Average: 100.52 Mbps, standard deviation: 0.05 Mbps
  9. Rotterdam - Delta Fiber Nederland 101.37, 101.34, 101.40, 101.39 and 101.35 Mbps. Average: 101.37 Mbps, standard deviation: 0.03 Mbps
  10. Den Haag - Glasnet 101.93, 101.99, 101.95, 101.97 and 101.92 Mbps. Average: 101.95 Mbps, standard deviation: 0.03 Mbps
 

Conclusions

Of these 10 local test servers:

  1. 6 test servers give speeds of circa 100.5 Mbps
  2. 1 test servers give speeds of circa 101.4 Mbps
  3. 3 test servers give speeds of circa 101.9 Mbps

Hence the speed that you measure depends on the testserver used.

We cannot properly explain these results and are puzzled.

This is mainly because we always assumed that your true internet speed is 12% more than the advertised speed.

However, if you assume that the Ookla speed test measures 90% of the maximum bandwidth and we measure 101.9 Mbps, then we have to conclude that our internet service provider offers 101.9 / 0.9 = 113.22 Mbps, while we have a 100 Mbps subscription.

That is 13.2% more instead of 12% more. We currently have no explanation or that extra 1.2%. If anyone can explain it? Please do.