[SOLVED] Understanding speed tests

tess38654

Reputable
Mar 10, 2017
5
0
4,510
We live in a condo and have no choice on our internet service provider (we used to have choice, and we had and individual contract with comcast and were quite happy, but the board decided to provide internet to all and the new provider disconnected our comcast).

Anyway, service is awful. Every time we get a service appt we are told our service is actually great (ha!) because they run a speedtest that shows how great they are.:rolleyes: They also use this same service to convince, repeatedly, our board that they are meeting service requirements.

But, it's horrible. I know just enough about networking to be slightly more than ignorant. ;-) We do need consistent connection for work, and that simply doesn't exist anymore (and neither of us have computers with ethernet connectivity, bleh!). However, we're using ethernet connection for our TVs, but even that is spotty - trying to stream a movie at night is horribly frustrating. Basically, we can't have multiple devices utilizing bandwidth at the same time, so we are constantly censoring our usage, which is not ideal, especially working remote.

I ran 4 basic speedtests -testmynet, google, fast.com and speedtest.net (service provider utilizes this one). As you can see in the screen shots, the first three show approx the same result - speedtest.net is the outlier, but also, interestingly, speedtest.net is defaulting to a direct connection TO our service provider (Network Communications). If I change the server on speedtest.net, then I get results in line with the other three (11.7 this morning on download - not shown in the screen shots, though.) Also note, I ran these sitting about 4 feet from the router. If I move, it drops to about 2-5 Mbps. We do have two separate routers - one for upstairs and one for downstairs. Both have similar speeds, though.

When I mentioned this to our rep I was told that the only true test was using speedtest with their server (and they've told some board members and our GM this as well, which makes it difficult for me to debate since this is the narrative everyone is using).

So, can someone help me with my ignorance on this? I don't know enough to credibly argue, but intuitively, this seems like total BS to me? Our connectivity isn't always routing to "their server", right? So seems to me that a true measure of speed shouldn't be predicated on just that. Or am I wrong? Help!

Thanks!
Inked-My-Post-1-LI.jpg
 
Solution
You need to always test with ethernet or you are going to get inconsistent results.

So lets take a fake example. The latency of 16ms on your test shows that the server is farther away and not this simple.

So lets say I am scumbag ISP. I contract out for service to a apartment complex. I then connect all the apartments with ethernet cables so they can get gigabit service. I also place a speedtest server in the building connected directly to this same network. Now since the goal is make as much money as possible I only buy 100mbps connection to other larger ISP. So now I will sell you a gigabit service and prove you can get it by testing to my local server. When you test to other internet providers you will only get...
You need to always test with ethernet or you are going to get inconsistent results.

So lets take a fake example. The latency of 16ms on your test shows that the server is farther away and not this simple.

So lets say I am scumbag ISP. I contract out for service to a apartment complex. I then connect all the apartments with ethernet cables so they can get gigabit service. I also place a speedtest server in the building connected directly to this same network. Now since the goal is make as much money as possible I only buy 100mbps connection to other larger ISP. So now I will sell you a gigabit service and prove you can get it by testing to my local server. When you test to other internet providers you will only get 100mbps. The ISP will correctly say that have no control over traffic on other ISP. They will attempt to blame them ignoring that the other ISP is delivering exactly what they are being paid for.

You can try things like tracert etc to various sites to see what ISP the traffic passes through. Maybe you can find test servers on different ISP that run faster. In the end though you have no ability to change anything. Maybe a VPN service would help but you need to find a vpn service that you get fast speeds to their data center and hope they have a better path. Big problem will be finding that VPN data center that is not affected by whatever bandwidth cap you have.
 
Solution
No just always test exactly at the same spot very close to the router. If you can get 95mbps to the ISP server then you know the wifi is not the bottleneck. You just have to be a lot more careful when you test with a wifi connection. As you have found moving the device can cause a large change in the performance. You need to be careful that this is not a factor in your testing.
 
More the questions is even if you can find it what can you possibly do about it. So far what your test results show is if you attempt to go out your ISP network there is a limitation of 12mbps. You can try to get the ISP to show you any site that is not inside their network that gets better performance.

This is mostly a learning exercise you likely will not get anything fixed. You would need to figure out how your ISP connects to other ISP and if there is any difference in the speed. TRACERT to a lot of different addresses might give you a clue.

So as a place to start, you only really need to look at the diagram to start.
https://en.wikipedia.org/wiki/Tier_1_network

What is key here is the connectivity of the tier1 ISP is really the core to what makes the internet work. Big companies by directly from tier1 providers because of all the interconencts.

A company like comcast is a tier2 ISP. This is still a huge ISP that has very good connectivity to the tier 1 ISP.

What I suspect you have is a small tier3 ISP. What you hope is they are not a single home tier 3 ISP like is shown on the lower right. This type of ISP only has a single path to the internet....ie the tier1 providers. Any issues on this path will affect all customers. What you have to hope is there is more than 1 connection and you can somehow find that and use it.

In the end you are basically stuck. You have to get all the other owners to complain with you. Maybe your management company will do something.....assuming there is not something corrupt going on like they are getting kick backs from the ISP. At least try to get them to allow a second ISP into the building that you must pay for yourself.
 
  • Like
Reactions: tess38654

tess38654

Reputable
Mar 10, 2017
5
0
4,510
According to my state law, I have choice if I want it. I would like Comcast ;-) I know it worked for ten years for us (and as recently as 2017 when we were disconnected). I contacted Comcast, and they said they'd be happy to turn services back on, but apparently the other provider just disconnects it. Not sure if that's legal, but... they claim it causes interference.

However, while the law is clear that I can have choice, it's up to me to figure it out. My guess is I could exercise this right, but would cost me thousands of dollars (?) to have Comcast run a separate line for me (we're near the top of a high rise, plus they'd have to run new cable to wherever groundside I guess?)

All the owners are upset about this. Not just me. So much so that the board is finally going to change providers when the contract is over. Um....to Direct TV (ugh) who subcontracts out to White Sky for internet, which I doubt will be any better than current provider. That's not for at least another year, though, so...Yeah. I'm stuck. And I am trying to educate myself now because my guess is I'll still be fighting this when they change providers. Bleh.

Thanks for the help/information!!!!