7.1 legacy conversion. calling experts for help with drastic whole in my qos knowledge

why did i think i would just multiply the byte-count by 8 to get the BW to convert from CQ to MCQ?  and then apply the MQC's BW statements without the percentages as the SG has....(bandwidth 40000 ..and not bandwidth percentage 50)

...why are they talking about ratios in below link?  i have researched but still have to guess about the ratio part.  is it  because the CQ does a round robin type thing (this que go, now this que, now this que...now start over)...  so its like..totally different then the MCQ algorithm..

http://forum.internetworkexpert.com/ubbthreads.php/ubb/showflat/Number/23065/page/3#Post23065

 

 

Comments

  • <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">





    The values used in the custom queue don't map directly to bandwidth
    values.  It is a byte count that controls how many bytes are removed
    from that particular queue when it is serviced in a round robin
    fashion.  If you have two queues each with an equal byte count it means
    that they each get 50% of the bandwidth.  If you have one queue with a
    count of 1000 and a second with 2000, it means the first queue gets 33%
    and the second gets 66%.  The problem is that the ratio doesn't stay
    the same when you apply it to different speed links.  For example a
    custom queue applied to a DS3 link and to a FastEthernet link will not
    end up in the same amount of bandwidth reserved, but it will be the
    same ratio.  The MQC makes this config much clearer by simply saying
    "bandwidth xyz" where "xyz" is how much you want to reserve in Kbps.





    HTH,



    Brian McGahan, CCIE #8593
    (R&S/SP/Security)

    [email protected]



    Internetwork Expert, Inc.

    http://www.InternetworkExpert.com

    Toll Free: 877-224-8987 x 705

    Outside US: 775-826-4344 x 705

    Online Community: http://www.IEOC.com

    CCIE Blog: http://blog.internetworkexpert.com






    danielowhite wrote:

    why did i think i would just multiply the byte-count by 8 to get
    the BW to convert from CQ to MCQ?  and then apply the MQC's BW
    statements without the percentages as the SG has....(bandwidth 40000
    ..and not bandwidth percentage 50)

    ...why are they talking about ratios in below link?  i have
    researched but still have to guess about the ratio part.  is
    it  because the CQ does a round robin type thing (this que go, now this
    que, now this que...now start over)...  so its like..totally different
    then the MCQ algorithm..

    http://forum.internetworkexpert.com/ubbthreads.php/ubb/showflat/Number/23065/page/3#Post23065

     

     







    "Internetwork Expert - The Industry Leader in CCIE Preparation

    http://www.internetworkexpert.com



    Subscription information may be found at:

    http://www.ieoc.com/forums/ForumSubscriptions.aspx

    "
  • thanks Brian...i think i have it now..

  • I'm still not sure I understand the conversion then.  If there are 2 queues and 1 is 1000 and the other is 2000 then the first queue get 1/3 of the bandwidth and the second queue gets 2/3 of the bandwidth.  So in this instance Q1=5000 Q2=3000 and Q3=500.  Wouldn't that mean that Q1 gets ~59% of the bandwidth Q2 ~35% and Q3 ~6%?

    Is it correct to add the values together and then take the percentage of the whole for each queue? 

  • The queue size percentages are actually correct when you add in the default queue size of Q4 which is 1500.  Q1:Q2:Q3:Q4 = 50:30:5:15 = 100

  • This was posted a while ago however my question is how would we know that the default queue-size is 1500 in this case ?

  • This is what IOS 12.4 docu says about default byte count:

    http://www.cisco.com/en/US/docs/ios/qos/configuration/guide/config_cq_ps6350_TSD_Products_Configuration_Guide_Chapter.html

    The following example decreases queue list 9 from the default byte count of 1500 to 1400 for queue number 10:

    queue-list 9 queue 10 byte-count 1400
    
    
    
    
    So it seems like a custom queue always has a byte count of 1500, unless you change it with the command above.
    Therefore Legacy Custom queue 4 will have 1500, queue 1 5000 and so on. So the examples in the comments above make sense it all adds up to 10 000. This can then be used as % for MQC.
  • i though i was fine with my solution:

    conf t
    class-map WWW
    match protocol http

    class-map FTP
    match protocol ftp

    class-map TELNET
    match protocol telnet

    policy-map EX_CQ
    class WWW
    bandwidth percent 50
    queue-limit 30
    class FTP
    bandwidth percent 30
    queue-limit 20
    class TELNET
    bandwidth percent 5
    queue-limit 20
    class class-default
    bandwidth percent 15
    queue-limit 20

     

    but i did not considered at all the inner policy and the shaping.Even if i think that the net effect of the SG solution

    and mine are the same. can some brave man explain me why the shape-average is needed?

     

    WTH, my EX_CQ cannot be applied on a subinterface! [:'(]

    now the SG is clearer.

     

Sign In or Register to comment.