Buzzatronic

Customize server used for the speed test

9 posts in this topic

I'm finding the speed test to be somewhat unreliable.  For instance this morning I checked my latest speed test results and it's showing ~35Mb/s although my connection is a 250Mb/s connection.  In the last week Domotz has hit a max of 235Mb/s which is great.  However it seems I can't really rely on the historical data because the location of the test changes.  Usually it goes through "Metapeer" but this morning I noticed it was really low so I ran a manual test and it went through a server in Canada (not too far fetched since I'm in Seattle, but still probably not ideal).

 

To make sure it's not really my internet connection, I ran the OOKLA speed test built into my router and it returned 220Mb/s.  I ran the Domotz test manually again and it only got 45Mb/s.

 

Is there a way to configure exactly which server you want to use for speed testing to create a consistent history of your connection?  I ask because my internet provider offers rebates if you do not get the advertised speeds and I don't want to start complaining if I'm getting bad data from Domotz.

0

Share this post


Link to post
Share on other sites

For the Speed Test, we rely on perfSonar, a global test and measurement infrastructure composed of thousands of servers maintained and used by science institutions and universities  around the world.

In most locations they are comparable with ookla, but in others, they aren't, like I see in your case.  Many people raised same complaint as you did. For this reason we have decided to move to using ookla-based servers. This is currently under development and we are going to release it in production within a couple of weeks. 

There might still be differences, as every time it will be executed, will connect to the nearest free server.

We also plan to add, later, the possiblity of selecting the preferred test server. 

 

 

 

 

0

Share this post


Link to post
Share on other sites

Makes sense, thanks for the quick reply.  Looking forward to the update. :)

0

Share this post


Link to post
Share on other sites

I would look a dsl reports new speedtest. I seem to get awesome repeatable results with more info than ookla

0

Share this post


Link to post
Share on other sites

I actually prefer the PerfSonar system to be honest-

 

those of you saying you get repeatable results should consider what you are saying. Is it really your belief that a broadband trunk will have the same response at 6pm as it will at 4am. These are like Highways and there will always be traffic at certain points that will cause congestion and variances. If you are not seeing these it is likely that the test is not a legitimate implementation, and is perhaps skewed as may be the case in both OOkla and speedtest.net who are both receiving large amounts of advertising funds from a few large players. Now I would never insinuate that a company that survives off of advertising funds would ever skew reports about connectivity results from their key clients as that would be inappropriate and completely unheard of in life in general. Just in comparison  PerfSonar is an implementation of a truly pure and scientific Link quality and bandpass test that is directed towards and managed by purely scientific organizations that do not operate off  any type of advertising blood money. I will say I was amazed at the quality and predictable results of my tests run by OOKla on the Comcast network despite getting significantly contradictory findings in both Speedof.me and personal experience. Lastly Dsl reports does seem to have updated their speedtest and it appears to be a genuine implementation of a tool, however great care was taken to shield the operation from prying eyes so there is little way to know the true nuts and bolts of the operation. It is clear that most the test legs are not hosted by Dsl reports but thats not anything of much concern as who wants to own a data center just to provide a speed test. Besides the key is to vary your test vectors and targets to get clear picture of the issue, which again is a good reason to stick with PerfSonar. Ok so besides counting myself as a backyard scientist and having fun with DryIce I have no connection to PerfSonar I just like their scientific roots. Ok so thats my vote thanks for reading.

 

Last thought, If my doctor told me he saw a dark spot on my chest Xray while I would hope for a smudge if it appeared in  4 successive Xrays at other doctors I would stop blaming the Xray and start chemo --point being the indication is theirs a bad network issue in the Canadian border region, amazingly a tap point for certain state level players, why are we blaming perfsonar again and the customers not complaining to the Broadband providers???

2

Share this post


Link to post
Share on other sites

Good food for thought iJOE. Thanks. What I meant by repeatable was that with DSL Reports new tool is that it was consistent (when I chose the params and the servers). Using things like ookla etc - it was all over the map minute to minute. Thank you for your explanation.

0

Share this post


Link to post
Share on other sites

fully agreed with IJoe, and this is why we implemented PerfSonar(which is much less stable and full of servers that are not working... but when it works is very reliable).

But unfortunately 90% of our users are complaining about the significant differences they get as compared to speedtest.net and they would like to use that.

 

So we are moving to that direction. But I am not going to delete the perfsonar code (and hope at some point we can have both, which I think would be great for users to have two different measures). I do not think we have resources for having both soon. But it is definitely something that I am very willing to do

1

Share this post


Link to post
Share on other sites

I know this post is coming up 8 months from OP's original comment but I too would like the ability to manually choose/customize the server I'm connecting too.

The area that I live in just had fiber installed and many of the users I support would flip out if they paid for a Gig but only reported 120 up due to the restraints on the server sides ability to report higher numbers.

 

I found that http://www.dslreports.com/ gives you the ability to confirm your connection speed whether it's Gig/DSL/Cable/Etc to assist you in getting an accurate speed test.

0

Share this post


Link to post
Share on other sites

Hi,

 

I would also like ability to configure a specific server (when trouble-shooting with an ISP, I am often asked to do speed tests to specific server)

The ability to schedule a bit more frequently than 6-hourly would help get a better picture of bandwidth patterns (hourly at most).

 

Thanks

0

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now