Pelican Parts Forums

Pelican Parts Forums (http://forums.pelicanparts.com/)
-   Off Topic Discussions (http://forums.pelicanparts.com/off-topic-discussions/)
-   -   Test your internet speeds (http://forums.pelicanparts.com/off-topic-discussions/956748-test-your-internet-speeds.html)

red-beard 05-16-2017 07:46 PM

Home internet

http://www.speedtest.net/result/6303350576.png

beepbeep 05-16-2017 11:01 PM

Quote:

Originally Posted by reachme (Post 9589543)
There is a problem with these internet speed tests. A couple problems really which is why you are finding variability..

1. You are not measuring speed at all, you are measuring throughput capacity. The (propagation) speed of a signal over fiber is about the same as a signal over copper. Add in very necessary but comparably very slow devices to boost and route the signal along the way and it’s a wash.

You are mixing up things. The test is measuring speed and median roundtrip delay for ping packets to nearest test server. "Necessary devices to boost signal" are called Optical Amplifiers and add almost no delay and are not slow at all. Basically, they are loops of doped fibre pumped by laser. There is no conversion of signal...amplification is done in light domain.

https://en.wikipedia.org/wiki/Optical_amplifier

Routing does create some delay. Of course, delay will be proportional to amount of hops and physical (and logical) length of the route.

reachme 05-17-2017 05:17 AM

Not mixing up at all, I said (boost and route) as a collection of devices.
You cannot simply boost a signal all along unless you have a direct fiber-fiber connection between hosts as I said. Even then the signal would need regeneration for long hauls but that scenario we understand is irrelevant. (we do right?)

That means switches and routers along the path and the most significant delay wise (not just some delay but orders of massively greater)- routers on the path change signal delay from microseconds to milliseconds. That is forever in network terms, like GT3 racers having to stop each lap and wait for their number at DMV office. Add in 10 of those routers on the path and the variability and you see is completely up to routing algorithms, capacity, priority etc. Delay is far from proportional or deterministic as you say. What is the point of measuring a GT3 cup race with the DMV wait times thrown in?

Besides "ping" is ICMP control traffic, tiny and carrying no data which gets a different priority than data traffic so as I said it has little bearing on actual data traffic and not even talking about Natting or content inspection delays.

I don't know what loops of doped fiber pumped by laser but it sounds dirty. :) Loops in fiber are bad and degenerate signal as light bounces, not strengthen. It's all doped-I think you mean single mode fiber but you may be looking at the straightaway and forgetting about the DMV.

widgeon13 05-17-2017 05:34 AM

Quote:

Originally Posted by Don Ro (Post 9588607)
Century Link... :(
.
<a href="http://www.speedtest.net/my-result/6300159153"><img src="http://www.speedtest.net/result/6300159153.png" /></a>


Identical to my Frontier service, actually embarrassing.

wildthing 05-17-2017 06:03 PM

Still on WiFi but on a Mac:

http://www.speedtest.net/result/6305976884.png

Also this was while my daughter was streaming music...

pwd72s 05-17-2017 07:10 PM

84.3... 6.21 Comcast cable


All times are GMT -8. The time now is 01:24 PM.

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2025, vBulletin Solutions, Inc.
Search Engine Optimization by vBSEO 3.6.0
Copyright 2025 Pelican Parts, LLC - Posts may be archived for display on the Pelican Parts Website


DTO Garage Plus vBulletin Plugins by Drive Thru Online, Inc.