Home
 » ISP News » 
Sponsored Links

Ookla Accuses UK Broadband ISP of Manipulating its Internet Speedtests

Wednesday, Jan 27th, 2016 (10:39 am) - Score 8,808

Ookla, the company behind the popular Speedtest.net service, has accused a broadband ISP in the United Kingdom of “intentional manipulation” by “prioritizing port 8080” Internet traffic in order to return a better performance result for their subscribers. But the reality may be more complicated.

The ISP in question is a smaller provider called Pulse8 Broadband, which has been steadily growing in popularity thanks to its “no contract” approach to FTTC and ADSL based broadband services. However the ISP now finds itself at the centre of a debate over whether some broadband providers are actively manipulating Internet speed testing services.

The situation first came to light after an article on Myce.com (formerly CDFreaks.com), which normally focuses on matters related to digital storage products, spotted that a customer of Pulse8’s FTTC (“Fibre Broadband“) service appeared to be receiving a much more positive result via the ISPs own speedtester (here) than when using the independent TestMy service.

The speedtest on Pulse8’s website is powered by Ookla and upon further inspection Myce.com claimed that the ISP’s test appeared to be prioritising traffic to TCP port 8080, which over the years has been used for all sorts of services from HTTP webcache to multiplayer games and remote web cams.

By comparison most web browser (HTTP) based traffic is usually carried over TCP port 80 or 443 and some tests prefer to use that, although we’ve also seen testers use TCP port 8095 and various others. Ookla itself claims to make use of TCP port 80, although after a quick search we also note that they’re familiar with 8080 (here and here).

Never the less Myce noticed that tests run via the Pulse8 connection on ports 8080 and then 80 via the TestMy service were returning different results. On port 80 the customer would return a result of 4.6Mbps download and 5.7Mbps upload, which suddenly jumped to 6.8Mbps download and 8.2Mbps upload when using port 8080 (consistent over several tests).

Seán Byrne, Myce Author, said:

Straightaway, it’s pretty obvious that the port 8080 test is performing better than the plain HTTP based test. So then I asked him by e-mail to run a multithread test with a fixed 50MB block size and to repeat the multithread test by accessing the website over port 8080. He ran these with his laptop connected directly to his router with an Ethernet cable. By using the same 50MB block size and test server, this is the equivalent to testing the maximum speed of two cars on the same test strip over the same distance.”

The result for the multithread test returned a download speed of 6.5Mbps for port 80 and 38.6Mbps for port 8080. Quite a difference. “So it is obvious that either his ISP or the TalkTalk Wholesale network it resells has given port 8080 special treatment,” said Seán Byrne.

A similar issue was also discovered with another ISP in the USA (Ziggo) and some non-UK mobile operators. However it’s worth noting that the same problem was NOT observed when the tests were conducted on other UK ISPs (e.g. BT Broadband, Sky Broadband and TalkTalk), which all appeared to behave themselves.

However we feel that testing with a single customer’s connection isn’t really enough to support the accusations being made by Myce. Similarly we’re not told any details about the customer’s line, such as what their VDSL profile details are (crucial to establish some context).

Likewise no tests were performed with other speed testing services outside of those mentioned and no direct download tests were performed to test the connection in different environments. Furthermore we would have liked to see the tests performed on different computer hardware, both wired and WiFi etc. It’s also unclear what time of day the tests were performed as this too can be impacted by changes in traffic (peak vs off-peak periods).

In fact there could be any number of reasons why a specific line might return an unusual result (e.g. firewall rules, capacity related problems or traffic management measures etc.), which is why it’s important to test with more than a single customer’s connection. On the other hand the situation has attracted a strong response from Ookla, which seems to agree with Myce.

An Ookla Spokesperson said:

The accuracy of Speedtest results are paramount to Ookla, and we take seriously any attempt to manipulate those results. We have begun looking further into the ISP mentioned in your article, and it does appear that they prioritizing port 8080. Whenever we discover intentional manipulation we work to correct it as well as ensure it does not occur again in the future.

Speedtest does employ measures to prevent manipulation and have new mechanisms coming online soon to further combat this. Luckily, only a handful of the tens of thousands of ISPs worldwide attempt this. Regardless, we never want to see ISPs attempting to manipulate results; this only misleads customers to the actual performance they’re receiving.”

ISPreview.co.uk decided to get in contact with Pulse8 and the ISP responded to say that they were surprised by the news. According to Pulse8, the speedtest on their website uses an iFrame (takes the content from a remote server) to load from the http://www.supportal-test.co.uk site, “which our Back Haul Provider provides so we can record results against customers speedtests.” A quick check shows that the supportal-test.co.uk domain is indeed registered to TalkTalk.

Pulse8 claims that in the past couple of weeks they’ve actually had the opposite happen, with some of their FTTC customers allegedly complaining that the test was returning consistently slower performance than other speedtest sites and as yet they have not identified the cause. The ISP has promised to contact their provider in order to find out what is happening.

In the meantime we are continuing to investigate and have liaised with some other speed testing services to see if we can examine the problem on a wider scale, although no doubt if any manipulation has occurred then ISPs may be quick to stamp it out before we can get any results. Still we would urge caution in laying the blame at Pulse8’s feet as Myce have done, particularly as the test is run off a site owned by TalkTalk, which Pulse8 does not appear to control.

Tags:
Mark-Jackson
By Mark Jackson
Mark is a professional technology writer, IT consultant and computer engineer from Dorset (England), he also founded ISPreview in 1999 and enjoys analysing the latest telecoms and broadband developments. Find me on X (Twitter), Mastodon, Facebook and .
Search ISP News
Search ISP Listings
Search ISP Reviews

Comments are closed

Cheap BIG ISPs for 100Mbps+
Community Fibre UK ISP Logo
150Mbps
Gift: None
Virgin Media UK ISP Logo
Virgin Media £24.00
132Mbps
Gift: None
Shell Energy UK ISP Logo
Shell Energy £26.99
109Mbps
Gift: None
Plusnet UK ISP Logo
Plusnet £27.99
145Mbps
Gift: None
Zen Internet UK ISP Logo
Zen Internet £28.00 - 35.00
100Mbps
Gift: None
Large Availability | View All
Cheapest ISPs for 100Mbps+
Gigaclear UK ISP Logo
Gigaclear £15.00
150Mbps
Gift: None
YouFibre UK ISP Logo
YouFibre £19.99
150Mbps
Gift: None
Community Fibre UK ISP Logo
150Mbps
Gift: None
BeFibre UK ISP Logo
BeFibre £21.00
150Mbps
Gift: £25 Love2Shop Card
Hey! Broadband UK ISP Logo
150Mbps
Gift: None
Large Availability | View All
The Top 15 Category Tags
  1. FTTP (5473)
  2. BT (3505)
  3. Politics (2525)
  4. Openreach (2291)
  5. Business (2251)
  6. Building Digital UK (2234)
  7. FTTC (2041)
  8. Mobile Broadband (1961)
  9. Statistics (1780)
  10. 4G (1654)
  11. Virgin Media (1608)
  12. Ofcom Regulation (1451)
  13. Fibre Optic (1392)
  14. Wireless Internet (1386)
  15. FTTH (1381)

Helpful ISP Guides and Tips

Promotion
Sponsored

Copyright © 1999 to Present - ISPreview.co.uk - All Rights Reserved - Terms , Privacy and Cookie Policy , Links , Website Rules , Contact
Mastodon