1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Odd Bandwith Management behavior

Discussion in 'Sveasoft Firmware' started by FDM80, Feb 18, 2005.

  1. FDM80

    FDM80 Network Guru Member

    I recently went from a 768kbps/128kbps line to a 5Mbps/2Mbps line so obviously I've gotten around to trying to set some bandwith management all over again in Satori. I've seemed to have come across some funny behavior though. It seems like when certain numbers are used, the bandwith limitation is much larger than maybe it should be.

    Just to isolate the download side of the equation, I set my upload limit to be 3000Kbps so my upload is "untouched". I have stuck in values into the download with these results:

    Input : Resulting performance : Difference
    6000Kbps : 5000Kbps (full speed, as expected)[/marq]
    5500Kbps : 5000Kbps (full speed, as expected)
    5100Kbps : 4800Kbps : 300Kbps (slighty slower than full speed, as expected)
    5050Kbps : 3900Kbps : 1Mbps (much lower than I expected)
    5000Kbps : 3800Kbps : 1.2Mbps (much lower than I expected)
    4250Kbps : 3200Kbps : 1Mbps (85% of original speed, which is close to the percentage of how much you are supposed to use, 3200Kbps is much slower results than I am really aiming for)

    Now my real issue is that once I get very close to 5000Kbps and lower, it seems that it takes an incredibly large amount of speed off of my connection (much more than it should). Even though my advertised speed is 5Mbps down, my ISP actually over provisions and I am able to hit 5.1Mbps down on some speed tests. It seems like between 5050 and 5100 is where there is a magic value that for some reason takes a whole 1Mbps off my connection. Does anyone know how the math for this is done? Even though the general consensus is to input a value that is 85% of your advertised speed. I would like my "final speed" to be about 85% of my true bandwith.

    Just thought I'd share my issue. Seems odd that this is happening. My performance hits weren't this large (percentage wise) when I was using 768/128. I'd like to avoid trying to find the magic number between 5050 and 5100 that gives me what I want. I think it's odd that a 50Kbps difference in input can result in a whole 1Mbps difference in actual performance.

Share This Page