[Hidden-tech] Reliable, accurate, meaningful Internet speed test?

Robert Heller heller at deepsoft.com
Sun Oct 18 10:27:32 EDT 2015

At Sun, 18 Oct 2015 09:40:08 -0400 ussailis at shaysnet.com wrote:

> There are several hardware issues going on here.
> First NYC is a congested area, so the "speed" could easily be less. Your
> packets are going to a machine along with many others, takes time to
> separate all this, which can enter into the speed.
> Then there is the route your speed test packets traveled. Darpa-net was
> planned to be very robust in a nuclear war. Packets travel on whatever
> route is available at the time and where the net software places them. Your
> test to NYC, in theory, could have packets going to Europe and back, tho
> not likely. While those to Tx could have a more direct route.
> Finally there is a signal-to-noise issue than I run into with clients. You
> mentioned DSL. At my house, Verizon DSL is restricted to what can go over
> telephone lines. Verizon claims 3 Gbps, which I never see.

We are NOT using DSL -- we are over the MBI Middle Mile fiber.  I am only 
using the *DSLReports.com* speed test.

And I think you mean 3MBits for DSL, not 3GBits!  3 Gbps is not possible over 
copper phone lines.

> Now my wireless modem box claims 54 Gbps, which if it were one bit/Hz of
                                   54 MBits, I think you mean.
> bandwidth would not fit in the entire WiFi band. The FCC and other users
> might frown on this.

I am using a hardwired computer to a *dedicated* high-speed point-to-point
wireless link (across the town common) between the Library and the Police
Station. The connection to the MBI fiber is at the Police Station (there is a
reason for this that is neither here nor there). The reason we are NOT using
the Library's MBI connection is because MBI/Axia charges an arm and a leg for
*each* connection to their fiber, which the town cannot afford.

*I* have already done a test of the point-to-point wireless link and it is 
more than fast enough, given our MBI connection speed.

> So some fancy modulation schemes are used to cram more bits/Hz down the
> pipe. This requires a higher signal-to-noise ratio (S/N) on the line,
> and/or thru the air.
> Noise? Where's this noise? There is always noise in every communication
> system. It may be weak, but it is there.
> A higher S/N requires either more transmit power, or less range. Max
> transmit power is fixed, so there is either a range reduction, or the magic
> in the box reduces the bit rate to accommodate the range.
> Now what has all this to do with a wired system, or even a fiber optic
> system?
> Same rules apply. Excess noise, too many users, insufficient power, even
> sharing the available power among users, all cause a lower thru-put. 
> So is there a meaningful speed test? Probably if the receiving unit were
> next door and hard wired to your box.
> Jim Ussailis
> National Wireless, Inc
> PS If you want to wade thru it (make a big pot of coffee first) see "The
> Mathematical Theory of Communication," by Shannon & Weaver.
> Original email:
> -----------------
> From: Robert Heller heller at deepsoft.com
> Date: Sat, 17 Oct 2015 16:54:38 -0400
> To: hidden-discuss at mm01.tnrnet.com
> Subject: [Hidden-tech] Reliable, accurate, meaningful Internet speed test?
> Is there such a thing as a reliable, accurate, or meaningful Internet speed 
> test? 
> We are having some weirdness with trying to figure out just what speed our 
> Internet connection *really* is.  Our provider is supposed to be giving us 
> 20Mbits down and 20Mbits up.  It does not seem to be that and when we run 
> speed tests we get 'weird' results.
> Speakeasy using the *Dallas, TX* server says we are getting 12.37 down, and
> 18.93 up, but their *New York, NY* server says something completely
> different,
> 4.70 down and 18.93 up. What does that mean, really? Why is it faster using
> the rather distant Dallas server vs. the fairly close NYC server? Is
> Speakeasy's NYC server a '486? Or what? Or is there something randomly
> screwy
> with Speakeasy Flash code?
> DSLReports speed test is much better, reporting 16.16/17.7 megabit/second

Robert Heller             -- 978-544-6933
Deepwoods Software        -- Custom Software Services
http://www.deepsoft.com/  -- Linux Administration Services
heller at deepsoft.com       -- Webhosting Services


More information about the Hidden-discuss mailing list