Check out new solution for ISP:
Splynx Isp Framework solution for ISP with Radius and Mikrotik API support


QT shaping, Is that good?

  1. 10 years ago

    Hello, I would question the programmer in shaping Mikrobillu. I use the QT model with kind PCQ. I do not know why but the value Limit At the customer is calculated from the value of which is shown as a percentage for a given tariff rate thus guaranteed. But this is the wrong model QT because AT Limit to be the sum always with the same result of higher values ​​Parent Limit At. I am sending a screen where it is clearly seen that the sum of the Limit AT exceeds the upper limit of At. This may be under heavy load line problem, because the shaping actually stops working.

    -image-

    Attachments

  2. admin

    Jan 2014 Administrator

    Hi,
    what about screenshot of tariff from mikrobill (id - 2 )?
    and tab sectors from router configuration (in mikrobill)

  3. admin

    Jan 2014 Administrator

    And what correct value and why you think so ?

  4. Edited 10 years ago by lukas_okoun

    Sorry for the delay time

    For example, there should be the correct value 59 (number QT) * 2Mb = 118000 kb or vice versa, for example, if we want to keep Limit At 12,800 kb and each of the 59 QT number should be 12,800 kb / 59 = 217 kb this would was the correct number. The same error is also Limit At danny group tariffs for the Global Limit At.
    If I had to go to extremes and actually in Mikrobill would not ever enter any guaranteed value, this should always be calculated based on the total possible connectivity. Traffic prioritization could be solved using QT priorities. But this is no longer the programmer to decide.

    -image-

    Here I send from another competitive program that produces shaping QT and thus it should appear correctly. It is nice to see if there would be a full utilization of all QT that the speed beautifully stretched and would not come any shaping collapse

    -image-

    If you have additional questions I will be available, mainly my point is that your shaping a fair and worked so well.

    Attachments

  5. admin

    Jan 2014 Administrator

    Where is screenshot i asked ?
    ----
    what about screenshot of tariff from mikrobill (id - 2 )?
    and tab sectors from router configuration (in mikrobill)
    ----

  6. admin

    Jan 2014 Administrator
    Edited 10 years ago by admin

    Why you not ask why max-limit = 25600 not 236000 ?
    59 (users) * 4000 = 236 000 not 25 600 ...

  7. Do this screen?

    -image-

    ----
    Why you not ask why max-limit = 25600 not 236000 ?
    59 (users) * 4000 = 236 000 not 25 600 ...
    ----

    No max limit that's another story. But at the moment I'm interested Limit At because it is well-made shaping substantial.
    If the sum exceeds the max limit is subordinate to a superior, nothing major happens. However, if you exceed this limit by at such a big problem. It is perfectly logical.
    Deliberately simulate it if you do not believe me. Do you need 5 QT limit at 1Mb and 3Mb max limit a parent nadřezený do with 3Mb and 10Mb limit at max limit. Then all 5 QT put in a full load and you will see that it will be a problem.

    Attachments

  8. admin

    Jan 2014 Administrator
    Edited 10 years ago by admin

    The answer is agregation
    1:10 with 50% guaranted speed?

  9. But aggregation thus addressed by the scheme can not be a valid PCQ. It just is not possible to limit subdivisions at QT exceeds the limit value at the parent QT. Can not. If you want to aggregate so it makes little differently. Create another instance of QT = QT aggregate and then gives te 10 QT and QT when necessary 5 would be 1:5.
    At the limit should be calculated based on the total connectivity based on the number QT using the parameter limit at. If this is met, do not shaping problem at full utilization of the network. If you violate this so none of those users that do not reach the ceiling at her. It's perfectly logical. I do not know it and explain it :)

  10. admin

    Jan 2014 Administrator

    If you use aggregation 1:10 for normal use guaranted 10%... This will be better...

  11. I beg you to simulate that all QT fully charged will see what happens. If you set 1:10, and 10% low and back again in a number of burn out. By just one problem removal. So this deal is meaningless. Your do not want to train here somehow but this is just clearly written in HTB manual. Mikrotik HTB obviously employed.

    http://wiki.mikrotik.com/wiki/Manual:HTB

  12. admin

    Jan 2014 Administrator

    Ok, another solution - not use aggregation ...

  13. What do you mean? When aggregating a blank so it give me error. Aggregation must be completed. When you write down 1:1 ie aggregation is not so at the limit will not change, the problem still remains. I do not know why he seen such a fix to make it exactly according to HTB scheme. After all, it's a simple formula to calculate the limit at

  14. admin

    Jan 2014 Administrator
    Edited 10 years ago by admin

    make aggregation 1:1 and post screenshot here ...
    queue tree screenshot

  15. I set 1:1 and now it does not (at least not the parent NeoGroup_Main_2_DOWN), but I do not know how to maintain when there would need to add 10 more users to the same tariff. Only time will tell. However, if you add up the red and green rectangles so the result is more than the blue rectangle 145Mb. Which is still the same problem that I describe here and still not talking about that the tariff is about 10 and I limit all added up to be at much above the 145Mb.
    -image-

    Attachments

  16. Hello, you are the problem with limit at repair? I would like to hear some reassuring answer :)
    I'm not asking you to have it repaired immediately but at least I need to know what are you going to do about it.
    Otherwise, the exact wording of what I describe here can be found right here: http://wiki.mikrotik.com/wiki/Manual:HTB#Dual_Limitation

    I quote: Sum of committed rates of all children must be less or equal to the Amount of Traffic That is available to parent.
    and
    Maximal rate of any child must be less or equal to the maximal rate of the parent.

    Thank you for your response and time.

or to reply!