[DKTaber]

I have used Speakeasy.net for many years to measure the speed of my Comcast cable connection. Speakeasy gives you several servers to use for the test. I live in Delaware, and if I pick the NYC, DC or Atlanta servers, the speeds are unbelievable; for example, NYC a few minutes ago: 21Mbps download, 7.3Mbps. But if I select Seattle, the speeds are substantially lower: 10Mbps download, 1.3Mbps upload.

Why is there so much difference based on the location of the server? I have always assumed that it was because using a server very far away does not really measure the native connection speed because the data is having to travel through multiple servers or terminals between Delaware and Seattle. When I pick an eastern server, the packet travels through far fewer intermediate servers/terminals, so can get to and from the test server much faster. Is that true, or am I (as usual) all wet?

[cybertech]

I live in the northwest and find the same thing with the Seattle server. I use San Francisco. I like the results better!

[DKTaber]

Originally Posted by cybertech

I live in the northwest and find the same thing with the Seattle server. I use San Francisco. I like the results better!

Yeah, SF gives better results that Seattle. . . but not nearly as good as servers closer to me. The question is "Why?", and given the differences, how can these be accurate indications of the NATIVE connection speed (i.e., without the "interference" of intermediate servers or switching terminals)?

[cybertech]

I will move your post to Networking. That is where you will get the technical information you are requesting!

[DKTaber]

I wondered where to put the thread, but TSG doesn't make clear what's "Web" and what's "Networking". The Web is Networking, but Networking is not all Web! Would be helpful if TSG defined it better.

BTW, I have the same issue with the little icons in the first column next to thread lists. Some are padlocks, some are clocks, some are envelopes with messages, but when I open them, regardless of padlock, clock, envelope, they're closed. I could find nothing anywhere that explained what the icons mean. ?????

[cybertech]

I know that John Will travels this forum and I am sure he will be able to answer the questions if no one else does.

As for the icons...look at the bottom of the page, in any forum, below Filter & Sort threads for icon explinations.

[JohnWill]

Actually, the WEB is a specific subset of Internet access. While people tend to believe that the WEB is the Internet, that's not the case. The Internet was around from the early 80's, it wasn't until the mid 90's that the WWW came into being.

Some userful background information: History of the Internet.

Network encompasses a lot more than the Internet, though again, many people equate the two. Again, perhaps a definition of Networking might help.



To the specific question of Internet speeds, the farther away the host is, the slower in general the access will be. There are some "tweaks" that are possible given a specific set of circumstances to help with long distance speed. In general, if you're trying to test your local ISP speed, the closer the test hosts is, the better.

Here's my test to a local server.





Here's my test to a West Coast server:




Note that download speed is similar, it's the upload speed that is frequently more affected by distance.


Going to London slows me down a bit:





Another important thing to consider is the latency.


Ping and latency to a local server:





Ping and latency to a West Coast server:



Notice the difference in ping times to a local server and a distant server.

[DKTaber]

I really wonder how accurate these tests are. I went to Speedtest.net and did a down/upload test using the same Phila. server you used (you and I must live close to each other). If that server is really in downtown Philly, it's no more than 25 miles from my computer. The download speed was only a little over 8Mbps; upload was almost as much -- just over 7Mbps. I then tried the same test with a server across the river in Red Bank, NJ: 21Mbps download, 7Mbps upload. Did it twice on both servers, results were the essentially the same both times. So here we have a download speed that is 2.6x HIGHER from the Red Bank server, which is sustantially farther away than the Philly server. Go figure!

[JohnWill]

I've always found the tests to be fairly accurate, they certainly are very accurate for my connection. There will be variability depending on the exact path to the server and how busy it is. On the Internet, routing is everything when it comes to throughput.

[DKTaber]

Originally Posted by JohnWill

. . .There will be variability depending on the exact path to the server and how busy it is. On the Internet, routing is everything when it comes to throughput.

Apparently. So I go back to and expand upon my previous message. When you do a speed test, you are not measuring the intrinsic capability of your ISP, but the speed at which your connection received or delivered a file to a specific server under traffic conditions that existed at that moment. And "My download speed is XXMbps." is a false statement.

I'll mark this thread as solved.

[JohnWill]

Originally Posted by DKTaber

And "My download speed is XXMbps." is a false statement.

Well, that's a pretty strong statement which I don't agree with. Perhaps a slightly more correct statement would be: My download speed is at least xxMBPS, since it's been demonstrated to be that from one location.

[DKTaber]

Can't disagree with that. Perhaps I should have said "My download speed is XXMbps" is an incomplete statement. To have any meaning, it has to include the specific server and when it was run.

[JohnWill]

Correct, we're in total alignment now.

[Dezaras]

This is why the ISP's fine print almost always mentions "up to" XX mbps when selling you tiers of service for connection speed.

[DKTaber]

Originally Posted by Dezaras

This is why the ISP's fine print almost always mentions "up to" XX mbps when selling you tiers of service for connection speed.

Yeah, I was going to mention that. If you think about it, it means you could be paying for 6Mbps download speed and getting only 1, and they could say, "Well, we said 'up TO'". So it effectively means that anything over 0Mbps is OK. What they should be forced to say (by the Consumer Protection Administration) is "at least". Then, if you get less than that, they would have to fix it or reduce what you pay.

Thankfully, I'm on the other side of the spectrum. Most of my friends and business clients get slightly less than what they're paying for (e.g., paying for 3Mbps and getting 2.460 is the actual case of the small business my wife works for). I'm paying for 6Mbps, but even using the Seattle server on Speakeasy.net, I get 10Mbps most of the time, and have never gotten less than 8Mbps any time, anywhere. For servers on the East Coast, I seldom get < 16Mbps.