Jump to content


 


Register a free account to unlock additional features at BleepingComputer.com
Welcome to BleepingComputer, a free community where people like yourself come together to discuss and learn how to use their computers. Using the site is easy and fun. As a guest, you can browse and view the various discussions in the forums, but can not create a new topic or reply to an existing one unless you are logged in. Other benefits of registering an account are subscribing to topics and forums, creating a blog, and having no ads shown anywhere on the site.


Click here to Register a free account now! or read our Welcome Guide to learn how to use this site.

Photo

Wide variations in bandwidth speed tests among test sites


  • Please log in to reply
No replies to this topic

#1 Balsamea

Balsamea

  • Members
  • 1 posts
  • OFFLINE
  •  
  • Local time:01:04 AM

Posted 04 March 2013 - 01:42 PM

Reporting a problem to my ISP, which bandwidth speed tests are true/accurate?

 

Getting satellite internet speed tests all over the map, ranging from 2.88 Mbps to 21 Mbps (download speeds) using various test sites.  (Results in list below.)  Sites that give low results do it fairly consistently.  Sites that give high results also within a consistent range.  Same for ones that give mid-range results.  This includes some sites that use the same test engine (Ookla) -- yet they get widely different results.  Ran tests to many servers around the country, last night and today.

 

This will be a long report, because I believe full detail may help and will save us having to do some of the usual Q&A.

 

Running WindowsXP-SP3 on a 2009 Acer AspireOne Netbook (with all ports loaded with external devices to make it just like a real computer).  Machine runs great, always did, no problems except typical software hassles, but they've always been fully resolved.

 

I am running with a brand new Ethernet cable directly from the PC to the satellite modem.  No router.  Got tired of having to reboot it for downed WiFi.

 

Running Norton 360 security software, including their firewall, not Windows firewall.  Set up firewall to accept ping tests from certain servers, as advised somewhere reputable (I forget where).  I will turn that off later.

 

I have traceroute results.  Used it to get round-trip time for adjusting TcpWindow (buffer) size for speed.

TCP tuned for high speed according to advice

here: http://www.psc.edu/index.php/networking/641-tcp-tune#WindowsXP

and here: http://rdweb.cns.vt.edu/public/notes/win2k-tcpip.htm

 

Provider: DISH Satellite Internet-only, no TV package (uses Wildblue / become Excede / became ViaStat, expecting Kraft Food next - snicker)

 

Service installed Feb 2012, within a month after they released high-speed satellite capability.  It's my only broadband access option ... boonies.

 

Originally they guaranteed 12Mbps speed - routinely got 14 to 18 during many tests in first several months.  Usually used AT&T speed test because it seemed to give the most consistent results and was not trying to sell me stuff.

 

A few months ago I got a letter from DISH saying that with ViaStat now in charge of my connection, they are dropping guaranteed speed to 10Mbps, and in 18 months will charge $10/month more for it!  Yay!  In Europe they pay half what we do for twice our best available speeds.  Capitalism, American style.

 

Various quirks (erratic times to do same things) with websites and more frequent email POP connection failures (okay after retry) led me to see if satellite was culprit.  My fave AT&T speed test not there any more.  Site says to use speedtest.net.  That gives me really low results ... 3 to 4 Mbps.  I mention the AT&T connection because they referred me to speedtest.net, as did DISH, as did broadband.gov.  And speedtest.net gives me the lowest test results.

 

Norton online backup crashes my satellite modem, and DISH says, yeah so don't do that.  Norton says it's impossible for it to be happening if my satellite connection is working right.  I abandoned Norton backup for now.  It is junk anyway.  Waiting for them to come up with a better one, because my package comes with 25Gb space.  Anyhow, this is another reason I started looking at bandwidth issues.

 

Called DISH tech disservice department yesterday about very low speed test results.  They essentially said what they usually say: SO?

 

After telling to to run speedtest.net and got bad results, they put me on hold, then came back and said to use broadband.gov test.  19 Mbps.  It uses Ookla test engine, as does speedtest.net.  Gov site also offers alternative test via M-Lab.  M-Lab test is broken, both at broadband.gov and measurementlab.net.

 

Asked DISH why such a big difference between speedtest.net and gov test, both using Ookla, and which should I believe?  Which do they believe?  LOL  They like the gov test.  Sure.  The FCC is a lobbying organization for the industry.  Why should I believe their test over several others that report differently?  DISH says dunno.

 

I demanded repair.  Okay, $95 to come here, regardless of whose equipment is at fault.  (Because I don't have the guaranteed uptime that they provide to people who have a TV package ... I have internet only.  After all ESPN certainly IS more important than Internet access.)  I told them I'd get back to them.  Not eager to pay $95 for a tech to say there's nothing wrong.  So, meanwhile during further research, even 3Mbps is rarely an issue for me because I rarely download or upload large stuff.  But I want what I'm paying for, and so that I can do big stuff when I need to, like online backup, etc.

 

Today I ran MANY tests at several sites, on several servers.  Carefully documented every step (screen shots) so that I can repeat it exactly in ongoing testing.

 

So, my primary questions for our best wizards here:

 

1.  What could cause such widely varied speed test results, tending to be consistently fast, slow or moderate depending on the test site used?

 

2.  Why would multiple test sites advertising that they are using the Ookla engine consistently come up with different results, but, as above, each site tends to be usually within a certain range?  E.G.: One site usually 2 - 5, another usually 6 - 10, another usually 12 - 20, and they stay that way over several hours of testing over two days.

 

3.  Are there a few test sites I should trust/believe accurate over the others, and why?  Ultimately I am looking for a way to prove whether DISH is providing what I'm paying for.  They won't accept my report that one or even two test sites give low results.  Ideally it should be three or more.

 

4.  Apparently there are multiple ways to implement the Ookla engine.  What's up with that?  Is there a way of factoring the results to put them all on the same playing field, depending on the type of implementation (if that can be discovered)?

 

Help, suggestions, advice, prayers, gifts, dancing girls all welcome.

THANK YOU !!!

 

Test results today:

All figures in Kb/sec. Different numbers for each test site are for different servers, where the option is offered.  Otherwise they are just repeat tests.

 

First number is download speed / second is upload speed.

  1. Broadband.gov (Ookla version):
    1. 20386 / 512 Kbits/sec
    2. 19908 / 507
    3. M-Lab version broken

 

  1. Speedtest.net (Ookla)
    1. 3570 / 390
    2. 10570 / 1930
    3. 3360 / 540
    4. 3520 / 540
    5. 2990 / 470
    6. 3030 / 480
    7. 3320 / 520

 

  1. M-Lab (measurementlab.net) - Broken

 

  1. Adobe Flash (dslreports.com)
    1. 7225 / 632
    2. 5736 / 695
    3. 6960 / 679
    4. 7746 / 600
    5. 5405 / 557
  1. AT&T (att.com/speedtest) - Down; recommends speedtest.net
  1. Speakeasy.net (Ookla)
    1. 13200 / 610
    2. 24210 / 2330
    3. Back to Broadband.gov (Ookla) 19378 / 601
    4. Back to Speedtest.net (Ookla) 3590 / 550

 

  1. Bandwidthplace.com
    1. 2880 / 730
    2. 7120 / 700
    3. 3600 / 730
    4. 4000 / 730

 

  1. TestMy.net
    1. 19200 / 689
    2. 21100 / 359

 

 

THANK YOU !!!



BC AdBot (Login to Remove)

 





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users