TechSpot

Question about accurate speed testing and Choke points

By Lightingbird
Mar 30, 2010
  1. I have a question about download speeds.

    There is a customer in my local area that has service with Charter communications. They are paying for 60mb down and 5mb up. They are not getting the speed they are paying for at all. A couple of techs have been out to their home to resolve this an stated this was a line problem. So that is still waiting to resolved. Here is the odd question. When trying to measure the speed I am getting different results.

    Here are all the average test results:

    These tests were all done with direct connections to a ubee modem.

    Speakeasy Speed Test
    http://www.speakeasy.net/speedtest/
    20mb down 5mb up

    Speedtest.net
    http://www.speedtest.net/
    15mb down and 5mb up

    Testmy.net
    http://www.testmy.net/tools/test/d_load.php
    11.5 down and 5mb up

    Toast.net
    http://performance.toast.net/default.asp
    4mb! down and 5mb up

    Charter communications (charters own speed test)
    http://spt01olvemo.stls.mo.charter.com/
    33.5 down and 5up


    So every test got the upload speed right but every speed test was different. Now all of the closest servers were selected but still this odd result. Now I know you should use your ISP's speed test but this can be misleading to the average consumer.

    So I'm looking for some opinions on the best most accurate way to test a customers download speed. Also is it possible for a routers firewall to choke and hurt the speed coming into a customers home?

    Any other idea's?
     
  2. poertner_1274

    poertner_1274 secroF laicepS topShceT Posts: 4,172

    I can tell you that depending on the time of day and amount of traffic the results will vary. I bet if you checked at 3am your results might be closer to the advertised 60mb. Remember 60 is maximum, not optimal.
     
  3. Lightingbird

    Lightingbird TS Rookie Topic Starter Posts: 44

    Thats true but I'm more concerned on how all the speed sites are much lower than the isp's. Its almost like the isp is doctoring their own test.
     
  4. tipstir

    tipstir TS Ambassador Posts: 2,387   +106

    This site test out speeds "<--"http://www.pcwelt.de/ashampoo-speedtest_/"->" I've disable link though you can copy and past it out. Speed ranges varies at the ISP level.
     
  5. Lightingbird

    Lightingbird TS Rookie Topic Starter Posts: 44


    I'm sorry but I have already tested 5 different speed sites and another really isn't the solution.
     
  6. tipstir

    tipstir TS Ambassador Posts: 2,387   +106

    You can test 5 and more and speed variations will be completely different. ISP gives you a rated speed for download and upload. Some offer a boost feature to increase it. But that boost varies too. Technology is always growing and the heavy demands from all of us here who have power full systems that can handle such speeds.

    You're first post is saying ISP rated speeds are not what the client is getting thus you had measured that by using speed measuring sites. I just gave you one more to test. Their ISP should offer such a test site to determine what speed from the ISP they're actually getting?

    Let those techs come out with their $5,000 line tester meter and see where the issue is? Bad digital connect ends can screw-up the connections, bad or old coax line, weak signal coming in. and faulty modem not getting the.

    What does the modem read? You can do that yourself: xxx.xxx.100.1 from any browser:
    Look for:

    Receive Power Level
    -x.x dBmV

    Transmit Power Level
    xx.x dBmV

    Cable Modem Status
    Operational

    DSL, Cable an Fiber will vary in speeds. Best connection is direct one. What I mean if they're not using the all-in-one modem where coax and phone for VOIP is being use. Throughput on those tends to be slower if you were to bypass that setup and use a coax modem only and loop tap connection from module box for the phone. This type of connection will yield higher throughput. I should know this is how my setup is here. In my other home I couldn't get Comcast to give me what I get from Cox here. Their answer was we don't offer that type of setup.
     
  7. Lightingbird

    Lightingbird TS Rookie Topic Starter Posts: 44

    Tip thanks for the replies but if you read my post again, the customers isp already identified it as a problem and a line tech is going to resolve it. Haha you posted all of that and I already stated it was a line problem.

    My main question of this post was to determine what is the best way to measure a customers download speed. All of those websites were giving different speeds but their isp speed test was the closet.

    Either way, someone helped me figure out the best way to determine this.

    Thanks for trying anyway.

    In the future, read the post. Haha no offense just saying.
     
  8. tipstir

    tipstir TS Ambassador Posts: 2,387   +106

    I've read the post I am just stating the facts.. Nothing is perfect speeds will vary.
    You're in the business you need to spot check them and find out what the results are and go with that. Speeds posted here from you going to vary per test. They're not 100% so don't go by that. Choke or not that's it unless your client wants to invest or which ISP. Anyway you see to know already what to do.

    Good luck!
     
  9. compdata

    compdata TechSpot Paladin Posts: 526

    That is quite a difference between the ISP site and the others. It makes sense to me that the ISP is going to be the highest as it is the most direct connection, however they do have the highest incentive to doctor the results.

    It is always difficult to measure an exact download numbers particularly at the high end because it can depend a lot on the connections between the measurement sites, the time of day (traffic load), how fast the computer itself is, etc. . . I doubt the router itself (unless it is very old) is affecting the speed - it is more likely the computer then the router. I noticed a big improvement going from an old P4 to a core duo system just because my computer itself was having trouble keeping up with all the incoming traffic.

    The numbers the ISP gives are also always going to be the theoretical max, which you are never going to achieve. Unfortunately this sets up the customers for disappointment when they don't see what they were expecting. I am usually happy if i am within 60-70% of the max rate. For example i just got our church connection upgraded to 25M/15M Fios and i measured a connection at ~19M/12M which i was pretty happy with given we were coming from DSL.
     
  10. compdata

    compdata TechSpot Paladin Posts: 526

    Also note that Cable Internet is a shared connection to a local hub so the connection gets hurt a lot more depending on other local subscribers usage (exact details depend heavily on the particular implementation). Fios is direct to a local hub then shared (over a wider bandwidth connection) to the Verizon office so my speed doesn't depend anywhere near as much on other users. When i had cable internet there were definitely times of the day it would slow way way down (evening was the worst).
     
  11. Lightingbird

    Lightingbird TS Rookie Topic Starter Posts: 44

    I get what you are saying and it makes sense.

    I have a new post up about this.

    Three of my customers computers behind the router get between 59mb to 61mbs on the test. So they are getting the speed paid for 100%. Although that's using the customers isp of course. One of the computers is only getting about half of that. I'm just about narrowed it down to a bad nic card in the laptop.
     
  12. compdata

    compdata TechSpot Paladin Posts: 526

    Good to hear that they are getting what they are paying for :) The main thing is that they be happy with what they get right?
     
  13. SNGX1275

    SNGX1275 TS Forces Special Posts: 10,714   +397

    You should do a tracert to those speed tests. The ISP speed test will be with the fewest hops so it has the least chance of having any speed hangups.

    For example, speedtest from my ISP takes 5 hops.

    It takes me 9 hops to get to speedtest.propelwireless.com in Kansas City, and the first real hop is identical for both tests. A hangup on any hop and the speed is going to go down (obviously). So your ISP is going to provide the best results because it is the closest, even if you choose closest server on other tests they still won't be as close as your ISP, and therefore they have a greater chance of any delays.

    Also, for a statistically viable test you will need more than a sample set of 1.
     
  14. captaincranky

    captaincranky TechSpot Addict Posts: 11,684   +1,877

    Just a word of advice, keep making replies like this, and you won't have any replies to bad mouth.
     
  15. Lightingbird

    Lightingbird TS Rookie Topic Starter Posts: 44

    Thats your opinion.

    There is nothing wrong with stressing the information received was already attempted or off topic.

    It's not meant to be offensive and shouldn't be taken that way.

    If your that sensitive, I don't know what to tell you.
     
  16. am2567323

    am2567323 TS Rookie

    fast

    dsl is faster imo
     
  17. jobeard

    jobeard TS Ambassador Posts: 9,315   +618

    let's clarify the testing environment

    the customer has ONE isp connection

    the site selected has Another ISP

    issue: each ISP must manage the traffic to give ALL customers on a specific gateway 'fair bandwidth'. All it takes to impact the point-in-time performance measurement is either end having a workload that requires thruput adjustments to avoid saturation -- and by gosh they sure will -- wouldn't you?

    IMO, the only bandwidth you can measure with any consistency is that from your router to the ISP Gateway router -- there's nothing between these two points and the variation in timings WILL be directly related to the issue noted
     
Topic Status:
Not open for further replies.

Similar Topics

Add New Comment

You need to be a member to leave a comment. Join thousands of tech enthusiasts and participate.
TechSpot Account You may also...