testmy.net always 50% less on upload than everywhere else

[pingram3541]

I'm subscribed to a 150/150Mbps FiOS plan.  Been using it for years with no issues (Verizon FiOS).

 

I'm trying to prove to my new ISP (Frontier FiOS) that ever since the change I've been experiencing bandwidth issues. Sometimes I've tested w/ results less than 1Mb (even when using Frontier's own speed test tool).

 

Over the last few weeks it seems to be sustaining around 90-100Mbps up/down which is still 1/3 less than my subscription, however when I test using this service, the upload always comes back around 50Mbps and the download is always around 100Mbps like the other sites are reporting.  The other sites are not all sing Ookla, some were java based, others used http file transfer via browser similar to this site.  I'm clearly not seeing any results close to 150 like I used to get but I feel the ISP will fall back on that "speeds up to" nonsense.  The big question here is why does testmy.net only show half the upload speed when no one else does?  This would lead me to conclude testmy.net might not be capable of giving me real results without some further explanation.  Hoping to find that out now if you don't mind thank you.

https://testmy.net/compID/679530124573

[mudmanc4]

Hey @pingram3541 ,

 The first thing I might suggest is to run quite a few more tests, on various test servers from testmy.net. This way we have a bit more targeted data to scour through. You can change the test server at the top of the home page, and or the test page.

 

I would also suggest running several 24 hour tests, on various servers. This will show differences between high and low bandwidth times per hours tested.

[pingram3541]

Thx @mudmanc4,

I realized I wasn't logged in when I ran the other tests and have since leased a new IP as well.  Thanks for suggesting the 24 hour test as I hadn't noticed that option and was just running random tests against random servers which over the last few days have all been spot on 100/50 which is what led me to this post.  Gonna try what you suggested and report back thanks.

 

Added note:  My edgerouter also shows real time throughput from my device edge (per MAC/per interface) and it also shows 100/100 when I'm running other speedtest tools and then shows 100/50 when I run TMN tests.  This is also comparing the WAN interface and my PC's MAC, both are identical within k's.  It now makes me curious about throttling so I'm now curious to see the route info during these tests.

[Pgoodwin1]

I didn't find any upload test data for you. I guess that's when you weren't logged in.

 

I typically get 5.8 Mbps up on my Time Warner Cable 50 down 5 up plan. 

[mudmanc4]

6 hours ago, Pgoodwin1 said:

I didn't find any upload test data for you. I guess that's when you weren't logged in.

 

I typically get 5.8 Mbps up on my Time Warner Cable 50 down 5 up plan. 

He has a few tests which were done without being logged in

[pingram3541]

I was having some issues completing tests when I had the SF test server chosen.  Many times the download would never start and would stay at 0% until I reset the size but then the upload would never complete.  I switched to other test servers and completed many tests before coming back to SF and then they completed just fine.  The results are now pretty much all over the place never exceeding 100mb in either direction and consistently never symmetrical.

[mudmanc4]

@pingram3541 , Testmy.net is not designed to grant the users theoretical data throughput capacity (or what the ISP claims is the highest level of data movement) , however more so to give an analysis in terms of network weakness from the modem ( or last mile) to the testing endpoint, exploiting network weaknesses in the route.

 

Testmy.nets testing servers are located at some of the most widely utilized and popular website / datacenters, this gives the user the tools necessary to again analyze at which point the ISP peers are either overloaded, or simply weak.

 

The more tests you take to different area's, the the larger the scope becomes, over timed tests, as well as overall performance.

 

Happy testing

[pingram3541]

Yes indeed and I'm hoping the 24 hours tests will help answers those questions.  The big one being, why does everywhere else show consistently symmetrical results even though they are always 1/3 slower than my subscribed plan, why does TMN consistently show non-symmetrical results?  That might also answer why sometimes skype is reporting that I am the problem side of a bad connection but speed tests will come back with plenty...there's a bad hop and maybe its a popular one.

[mudmanc4]

10 minutes ago, pingram3541 said:

Yes indeed and I'm hoping the 24 hours tests will help answers those questions.  The big one being, why does everywhere else show consistently symmetrical results even though they are always 1/3 slower than my subscribed plan, why does TMN consistently show non-symmetrical results?  That might also answer why sometimes skype is reporting that I am the problem side of a bad connection but speed tests will come back with plenty...there's a bad hop and maybe its a popular one.

What I can say about several other sites testing servers, some of them are located on the top tier 'channel' of the ISP's network, where little to no real world traffic moves through. 

 

Top that with their default settings utilizing port 8080, which again, is not the default HTTP port, so very little to no traffic moves over those lines. Giving test results in this manner, in my opinion, is good for internal network throughput only. Not necessarily real world is it. As websites and CDN's are scattered all around the planet, not necessarily on the big pipe with little usage. Which of course by logic will give a more consistent reading, considering.

 

 

[Pgoodwin1]

I typically run three combined tests in succession to get an average. I usually do that using two different TestMy servers. I also try to run tests at a similar time of day. After some period of days or weeks, you can establish what your baseline "normal" speeds are. Then you can look for trends over time. 

[pingram3541]

Thanks gents.  I finished the first batch of 24hr tests to Los Angeles and consistently see 90/50 gonna try another test location.  Now how do I go back on this test history and scrutinize detailed info such as hops?  Is that possible w/ TMN?

[mudmanc4]

7 hours ago, pingram3541 said:

Thanks gents.  I finished the first batch of 24hr tests to Los Angeles and consistently see 90/50 gonna try another test location.  Now how do I go back on this test history and scrutinize detailed info such as hops?  Is that possible w/ TMN?

This is a very good question, I like the idea of it. And as well I'm sorry to say I am not aware of the possible hop count being in the user database. It's possible @CA3LE has something going along these lines. I'm sure he'll chime in on this when there is a moment to address. 

[pingram3541]

Ok, the results are in and it still looks pretty consistent @ half the upload speed when testing with TMN.  I tested 3 separate 24 hour test between 2 different test servers and I kept them fairly close to home Los Angeles and Dallas.  Not sure how to get more detailed information response times per hop but heres what I feel confident in my findings:

 

- Not a single test, even my ISP's, ever show near my subscribed speed of 150/150Mbps

- Test w/ TMN consistently result in 2/3 subscribed speed download, 1/3 upload

- Everywhere not TMN, result in 2/3 subscribed speed both download and upload

 

Would love to see if we could break down the hows/whys on those 3 conclusions but I feel I can safely say its time to open one of those dreadful and often lengthy support tickets in which I'll probably get 2 new routers by the end of it all even after telling them I've swapped routers and even taken them to another connection proving they support upwards of 600Mbps throughput.  The most frustrating part of all of this is trying to explain how this stuff works to the folks in the tech support dept. as the first 2 tiers have a general knowledge at best and when you say something like "Do you see my router's MAC in the upstream arp cache?" and they'll glaze over.  I spent upwards of 2 hours trying to get them to release an IP lease for heavens sake.  Ok, end rant.  Good vibes, I still have decent bandwidth so I regress back to moderate bliss.