rvdsabu4life6
@lhtot.com
| What do speed tests actually measure
I have never ever trusted speed tests in my life because I have proved many times how inaccurate they are.
Well my ISP monopoly just put in our crappy fiber to our business. They had us test the speed. It was 2mbps down (the highest they offer). Downloading from ANYTHING is not more then 200kps. We have complained and they keep showing us the speed test and saying too bad.
I am filing several complaints against this company. I would like to try to replicate what a speed test does without using one. I am not sure if it uses ports only or what.
Does anyone have any suggestions? |
|
McSummation
Mmmm, Zeebas Are Tastee.
Premium,MVM
join:2003-08-13
Round Rock, TX
 ·AT&T Southwest
| said by rvdsabu4life6 :
Downloading from ANYTHING is not more then 200kps. Are you sure this isn't 200 kBytes/second? That would be about 2 mbps (mega bits per second)
What speed tests have you run?
Speed tests generally download a file of a megabyte, or so, and measure the time it takes. Nothing more than that simple test. |
|
tschmidt
Premium,MVM
join:2000-11-12
Milford, NH
 ·Hollis Hosting
·Verizon Online DSL
·Fairpoint Communic..
1 edit | reply to rvdsabu4life6
There are two different types of speed test. One that verifies first-mile connection is working as advertised and end-to-end to end performance testing.
First step is to make sure first-mile is working as advertised.
Then do file-transfer end-to-end test. The ones on the BBR tools pages and Speedtest.net are accurate and popular. Try to pick a nearby test server. File transfer performance is affected by distance and hop count.
As McSummation  posted are you sure you are not confusing bits and bytes? 1-byte = 8-bits. Best case file transfer rate for a 2 Mbps (mega bits per second) connection is 250 kBps (kilo bytes per second). File transfer rate (payload) will always be less then raw speed of the connection. Depending on type of connection, communication overhead can be between 5-15%. In addition any impairments between you and remote server will affect performance.
When doing performance testing are you sure there are no other PCs on local network and no applications on test PC accessing the Internet during the test?
Might I ask how you "proved many times how inaccurate they are."
/tom |
|
dynodb
Premium,VIP
join:2004-04-21
Minneapolis, MN
| reply to rvdsabu4life6
Speed tests are frequently inaccurate- have seen it many times, with some tests working fine with one provider but completely off the mark with others. Those that test by downloading multiple files at once tend to do better in this regard.
Speed tests generally shouldn't be taken as an exact measurement- they're best used as a general guideline to see if you're close to where you should be or not. Try three different tests, and you'll likely get three different results.
Find a good server and download a large file- it's not that unusual to have an actual download go much faster than was indicated by the speed test. |
|
vaxvms
ferroequine fan
Premium
join:2005-03-01
Worcester, MA
 ·Charter Pipeline
| reply to rvdsabu4life6
said by rvdsabu4life6 :
I have never ever trusted speed tests in my life because I have proved many times how inaccurate they are.
just wondering...
What have you done many times to prove the test are inaccurate?
--
If you think you are leading and no one is following you are just walking.
|
|
Omega
Displaced Ohioan
Premium
join:2002-07-30
Cheyenne, WY
clubs:  | reply to rvdsabu4life6
Sounds like someone doesn't know the difference between kilobytes a second and megabits a second.
--
Whats smells like blue? |
|
pog
Premium
join:2004-06-03
Kihei, HI
·Hawaiian Telcom
| reply to rvdsabu4life6
Speed tests generally measure average throughput. Literally, data is sent to and from the server hosting the test to your computer over an interval of time. Then, the amount of data is divided by the amount of time.... resulting in the amount of data per unit time.
I'm not sure what you mean by using "ports only"... but you can't really replicate a speed test without measuring speed. What you can do, however, is localize the test. IOW, add some software to your computer that will keep track of download/upload statistics for you. This will be better than any test site because it will be a true measure of the real world service you are getting from any site.
Maybe try »www.dumeter.com/ for the free 30 day trial?
--
My Site |
|
mr_slick
join:2003-05-22
Lynnwood, WA | reply to Omega
agreed... bits v. bytes |
|
Ray422
Premium
join:2002-03-04
Adger, AL
clubs:
·RoadRunner Cable
| I have wondered about this turbo-crap these ISP's are advertising. Isn't that just a momentary boost to fake (gimmick) the short speedtest, so they can say their speeds are better than actual?
Me wonders if a throughput test on a large file is the only thing that's reliable now days? |
|
mr_slick
join:2003-05-22
Lynnwood, WA
| that is what i always go by... i will usually fire up my news reader or download a windows service pack from microsoft and measure it with an app called "netpersec" as well as my firewall meter, and windows explorer messages. usually they will all agree (or be close). you have to account for overhead when using meters like netpersec or others, too. |
|
rvdsabu4life6
@comcast.net
| reply to rvdsabu4life6
that is what i always go by... i will usually fire up my news reader or download a windows service pack from microsoft and measure it with an app called "netpersec" as well as my firewall meter, and windows explorer messages
I do the exact same thing.
As far as bits vs bytes, I do not dispute that. However, I go by the actual numbers.
For instance, our DSL 256k connection got ~300-400kps on speed tests. In real time (newsreader, Microsoft download) we got 250kps sustained. At my house with Comcast 16mbps down, I get ~3mbps-4mbps actual. On Verizon FIOS on a partner company with 10mbps down, they get 2mbps down.
On our actual speeds, within 15 seconds, they go from 2mbps down actual to 200kbps down (slower then our old DSL). This is on every possible connection. It acts similar to Comcast speed boost.
I had high hopes that our business, which desperatly needs higher speeds, would benefit in some way. Now we seem to be paying double the price for less speed. |
|
rvdsabu4life6
@comcast.net | Ok so I was wrong. I Was indeed confused about bits vs bytes. That makes sense now in 50% of past situations.
Now my crappy ISP. When we were paying for 256Kbps, we would get 256Kps. Why then with fiber do we get 2Mbps for 100kps. |
|
SmackWeasel
join:2008-01-02
| reply to Ray422
said by Ray422 :I have wondered about this turbo-crap these ISP's are advertising. Isn't that just a momentary boost to fake (gimmick) the short speedtest, so they can say their speeds are better than actual? Me wonders if a throughput test on a large file is the only thing that's reliable now days? Indeed so Ray, My ISP uses "speed boost". On a speed test it shows up to 30Mb/s from my lowly 10Mb connection.
--
0111000001100101011000010110001101100101 |
|
Jerm
join:2000-04-10
Richland, WA
| reply to rvdsabu4life6
said by rvdsabu4life6 :
Ok so I was wrong. I Was indeed confused about bits vs bytes... Why then with fiber do we get 2Mbps for 100kps.
*sigh* rvdsabu4life6 please do some more research. 100 kilo per second? ROTFLOL.
2Mbps should allow for downloads around 250KB/sec max. With overhead probably a little less.
Sometimes the server you are downloading from is slow, or the transfer has issues (ie it will be 200KBps one second then 50KBps the next).
Do as others have said here, and use reliable speed tests (I personally like »www.speakeasy.net/speedtest/ ) and then track the actual second-by-second speed with an app like DUMeter. |
|
homerscousin
join:2009-07-04
| reply to rvdsabu4life6
Wow! I am surprised a fiber connection is so slow. I have done several tests at speedtest.net and others and also wondered how accurate the readings are. Just for verification and experimentation over a few days I went to a few file share sites like rapid share etc. and even Pirate Bay to download any old file that was BIG and watch the speed results. Also downloaded a 200 Mb file from Microsoft. Even my worst speedtest.net result was faster than any of these others and the best of the others was 1.8 Mb/sec (megabytes). I am not sure if this is what you mean by replicating a test but this kind of proved to me that any speed test will be on the high side. |
|
