Unusual findings.......

[BluJay614]

Sit down, grab a drink and a snack, this is a bit of a ramble that has a point with a question at the end.

Okay, so I finally figured out how to fix my spazmatic ethernet at home, even with it just having my laptop hooked up to the modem the internet kept dropping. I ended up picking up a new coaxial cable and Cat6 ethernet cable. Swapping the coaxial cable from the one my ISP (Spectrum) provided and making sure it was secure (thanks to an adjustable crescent wrench) to the modem. My ISP had provided a Cat5 cable, so for what was only supposed to be 100Mbps internet (IT WAS ALL THEY CLAIMED THEY HAD WHERE I LIVE!!!) should have been fine. Once I had gotten it working, I tried testing the difference between the two cables. I did speed tests on Spectrum's website, Google, Ookla (speedtest.net), and the Speedtest app on Windows 10 (again by Ookla). I kept getting some wierd results. Google kept wanting to use servers in Dallas, and kept giving me significantly slower speeds. Speedtest.com kept giving me numbers all over the place, some of which exceeded what I thought I should be getting. I honestly didn't do much testing on Spectrum's website. However, According to the Speedtest app, the three nearest servers(all 7 miles) are AT&T, Sprint, and "VTX 1".

AT&T's Ping kept bouncing between 25 and ~40, and the speeds where consistantly the worst. The download speeds wtih both cables where about 87.38Mbps and upload speeds where 5Mbps

Sprint's Ping stayed at right about 19 (dropped to 15 once), giving the lowest ping consistantly, with middle of the road speeds. Download speeds with both cables where about 90.89Mbps, Upload 12.5Mbps

VTX 1's Ping(30) and speeds seemed to be just, if not more consistant, with it giving the fastest speeds. Dowload speeds of 92.87Mbps, Upload 13.22Mbps

The only time it seemed to make a difference which cable was used was the upload with AT&T, where the Cat6 cable gave faster speeds, though it still never reached 8.25Mbps on that server.

ANYWHO, what I'm most curious about is if the Ping makes any kinda difference, and how was it that I was achieving such incredibly high downloads on speedtest.net?

[Firecheetah13]

Ping DOES make a difference, but not to the extent that you might believe. The longer the distance, the more spastic the results will be. Not necessarily across the board slower, but if you have ever played a competitive FPS like CS:GO, than you know that high ping causes hangups and stutters in game, especially since longer distances might have a ping that fluctuates. You have to remember that no two tests will be the same, and you have to account for some variance.

[brwainer]

A difference of less than 30ms between two servers is not noticeable, even when you are talking about low end results where 30ms might be 50% or 100% worse - it looks bad as a percentage difference, but it doesn't work that way.  Also, the "ping" results in speedtest.net aren't reliable, I don't know how exactly they are measuring that, but it almost always ends up significantly wrong, as in off by up to 50% in one direction or the other.

 

If your computer hangs a little, or overall doesn't perform fast, then the counter in the traditional speedtest.net test, which runs in Flash, can sometimes give incorrect results. The tests at beta.speedtest.net and dslreports.com/speedtest don't have this issue, as they are HTML5 based. Aside from that (which really should only come into play if you are using a netbook or very old computer), the package speed you pay your ISP for is rarely set as a hard cap, depending on their exact configuration. Most of the time, it is implemented in a way that if there is excess capacity available, then customers can get a bit more than what their package limits are. Of course this goes the other way too - if there is heavier usage in a neighborhood than the local equipment can sustain, then noone in that neighborhood will be able to reach their package amounts - hence why they say that speeds are never guaranteed.

[BluJay614]

22 minutes ago, brwainer said:

A difference of less than 30ms between two servers is not noticeable, even when you are talking about low end results where 30ms might be 50% or 100% worse - it looks bad as a percentage difference, but it doesn't work that way.  Also, the "ping" results in speedtest.net aren't reliable, I don't know how exactly they are measuring that, but it almost always ends up significantly wrong, as in off by up to 50% in one direction or the other.

 

If your computer hangs a little, or overall doesn't perform fast, then the counter in the traditional speedtest.net test, which runs in Flash, can sometimes give incorrect results. The tests at beta.speedtest.net and dslreports.com/speedtest don't have this issue, as they are HTML5 based. Aside from that (which really should only come into play if you are using a netbook or very old computer), the package speed you pay your ISP for is rarely set as a hard cap, depending on their exact configuration. Most of the time, it is implemented in a way that if there is excess capacity available, then customers can get a bit more than what their package limits are. Of course this goes the other way too - if there is heavier usage in a neighborhood than the local equipment can sustain, then noone in that neighborhood will be able to reach their package amounts - hence why they say that speeds are never guaranteed.

I feel like even on beta.speedtest.net is a bit buggy from all the adds that keep flipping around (No, I don't have an add blocker), which is something the app (looks just like the beta page of speedtest.net) doesn't have. dslreports.com/speedtest seems to be keeping things more consistant, showing the up to be ~12Mbps and the down to be in the low to mid 90's

[brwainer]

1 hour ago, BluJay614 said:

I feel like even on beta.speedtest.net is a bit buggy from all the adds that keep flipping around (No, I don't have an add blocker), which is something the app (looks just like the beta page of speedtest.net) doesn't have. dslreports.com/speedtest seems to be keeping things more consistant, showing the up to be ~12Mbps and the down to be in the low to mid 90's

I get what you mean about the ads on beta.speedtest.net. The DSLReports speedtest is different from others because it uses multiple streams at once - it should always be able to max out whatever your bottleneck is (normally your ISP connection/plan, but could also be wifi signals, or for some people an old router). Other tests like anything operated by Ookla (speedtest.net and even the Spectrum website you shared) are only a single stream, and sometimes individual connections get slowed down due to a router or link between you and the server that is malfunctioning or congested (router in this case meaning all the devices that "route" your traffic, including but not limited to the one in your house).