Thursday, 14 January 2010

Speedtest UPC Fiber Power 60 vs Alice DSL 20

I switched providers. I moved from Alice DSL to UPC Fiber Power. An explanation on the switch follows later, but here are some of the differences in broadband speed. Important disclaimer: This is not a perfect test. There are many factors involved.

The Networks
Alice is using ADSL2+ provided by BBNED in the Netherlands. This is an Unbundled Local Loop connection, so no KPN involved here. According to the Alice website I am close enough to the DSLAM to be able to get between 16 and 20 Mbit/s (screenshot).

UPC is using Docsis 3 to provide 60 mbit down and 6 mbit up to my home.

I tested using my laptop with the power plugged in. I shut down most software and processes that were running on the laptop. Especially the ones I suspected might influence the download. I shut down the wifi by flipping the physical switch on my T60 laptop. I used a 2 meter UTP cable straight to the UTP-port on the DSL-modem and Cable-Modem. There was no switch or anything in between. I used the Chrome browser and switched off all proxies and stuff. I used Chrome as I figured that with it's pre-loading of DNS queries it might be best and it consistently is the fastest browser around. It also has the least plug-ins installed.

I went to four websites to test the speed. Three performed speedtests and 1 allowed me to download a 100mb .bin file (104.857.600bytes according to windows). I used both UPC's and BBNED's speedtest sites and I always made at least two trial runs before performing three speedtests and for the file download I did at least one full trial file download before performing the test. The results are as follows (screenshots embedded below). I started at 23:30 and finished at 24:00. UPC was first and then Alice.

Results UPC
100MB File download from Leaseweb using the site: 29 seconds or 28926kbps
UPC Speedtest site: 40.87mbit/s down and 3.99mbit/s up, 49.51/3.64 and 50.47/3.99
BBNED Speedtest site: 29,433kbps down, 29,001kbps and 31,565kbps 10677kbps/2045kbps, 10144kbps/1735kbps and 13789kbps/2015kbps

Results Alice
100MB File download from Leaseweb using the site: 158 seconds or 5309kbps
UPC Speedtest site: 5.36mbit/s down and 0.87mbit/s up, 5.41/0.87 and 5.370.87
BBNED Speedtest site: 5363kbps, 5746kbps and 5762kbps 5222kbps/804kbps, 5276kbps/794kbps and 5269kbps/302kbps

Both networks don't reach the promised speeds all the way. What is shocking is that Alice doesn't even come close on its own network. It promises between 16 and 20Mbit/s but delivers only 5.7, regardless of what network you test against. That is only 28% at best. All in all switching to UPC got me 5.5 times more bandwidth for roughly the same price. (more on that later)

UPC fares better. To it has a problem it seems. It is only twice as fast there as Alice is. which would be an underperforming 17%. But when connecting to BBNed the worst I get is 48%. In it's own network UPC showed varying performance between 40 and 50mbps or a maximum of 84%.

There is more difference in the upload speeds it seems, but the tests probably aren't good enough to say much about it.


  1. Might be useful for follow-up tests: &

    Vlaai (SNT's OS ftp/http mirror) has evolved to a small cluster, serving multiple gigabits/s during larger releases. It should be a good benchmark for dutch providers with its SURFnet connection.

  2. We compared your results on our Ziggo 50 mb line, it seems that is limited to 10Mbps. I'm not sure if trackbacks work from wordpress blogs, so i'm commenting here just in case.

  3. As reference, I am Almere-Haven and just switched to KPN Glasvezel-silver. The cost is similar for TV,internet & phone lines.
    The interface on the KPN STB leaves some to be desired. I certainly miss BBC! The internet is fairly stable and speeds much better, with, as you say, symmetric dl/up coming soon.

    That said:
    Personally, I am very suspect of basing on such 'tests', for speed.
    I am currently on KPN Glasvezel (50 up, 5 down). I received vastly different results from compared to and others. (.nl being 10Mb slower than .net (>40 v <30)
    Since you are basing your comments on these services, I would be interested some explanation of how they actually 'work' and why the vast discrepancies.
    I enjoy your blog, even if I don't always understand all of it. Keep up the good work.

  4. Man, it's still surprising how retarded some people are. Can't you notice that takes 0,1 sec for it's tests. Which means the packet is too small before the transfer reaches it's peak.

    However, thanks for the write-up.


Note: only a member of this blog may post a comment.