CBWFQ 75% rule weird behavior

[durale1789]

Hi,

I got tested qos cbwfq using iperf and here is my configuration :

class ef

match ip dscp ef

class af21

match ip dscp af21

class af11

match ip dscp af11

policy-map test

class ef

  priority 512

   police 512000 conform-action transmit exceed drop

class af11

  bandwidth remaining  percent 50

class af21

bandwidth remaining percent 50

policy-map shape-test

class class-default

  shape average 1024000

  service-policy test

interface s0/0

bandwidth 1024

encapuslation frame-relay

max-reserved-bandwidth 75

service-policy output shape-test

In this configuration I should have for both classes af21 and af11 768-512=256/2 =128k each

However I get up  to 256k on each class af21 or af11 ... like if max-reserved-bandwidth was set  to 100% ...

I can t see any reason for such behavior ... are you ?

[Joseph W. Doherty]

Disclaimer

The   Author of this posting offers the information contained within this   posting without consideration and with the reader's understanding that   there's no implied or expressed suitability or fitness for any purpose.   Information provided is for informational purposes only and should not   be construed as rendering professional advice of any kind. Usage of  this  posting's information is solely at reader's own risk.

Liability Disclaimer

In   no event shall Author be liable for any damages whatsoever (including,   without limitation, damages for loss of use, data or profit) arising  out  of the use or inability to use the posting's information even if  Author  has been advised of the possibility of such damage.

Posting

Max-reserved-bandwidth doesn't control how CBWFQ actually shares bandwidth, only what class allocations you can define in pre-HQF IOS versions.

Since you shaped at 1 Mbps, and assuming your LLQ used its 512 Kbps, your af11 and af21 classes would split the remaining 512 Kbps evenly, i.e. the 256 Kbps you saw for each.  I.e. it's working correctly.

[durale1789]

Hi Joseph,

Since i m using ios 12.4(13)T, it should use HQF instead of pre-HQF. I understand your point to say that it is just a configuration limit so we can easily tweak it as needed up to 100%. However I don t understand why in books I see the following formula:

BW(available) = Max-res * BW - SUM(All BW guaranteed LLC; CBWFQ etc ...)

I have to test again CBWFQ without shaping because I think that since I configured shaping under the class class-default , the Hierarichal Qos may not  follow the 75% rule...

[Joseph W. Doherty]

Disclaimer

The   Author of this posting offers the information contained within this   posting without consideration and with the reader's understanding that   there's no implied or expressed suitability or fitness for any purpose.   Information provided is for informational purposes only and should not   be construed as rendering professional advice of any kind. Usage of  this  posting's information is solely at reader's own risk.

Liability Disclaimer

In   no event shall Author be liable for any damages whatsoever (including,   without limitation, damages for loss of use, data or profit) arising  out  of the use or inability to use the posting's information even if  Author  has been advised of the possibility of such damage.

Posting

Your 12.4(13)T should be pre-HQF, see: http://www.cisco.com/en/US/docs/ios/qos/configuration/guide/qos_frhqf_support.html#wp1118805

[smehrnia]

Hi,

did you assign you policy-map to ur interface? (coz i cant see that )

actually max-reserved-bandwidth only controls how much % of the bandwith is avalaible for allocation only, the rest depends on the flow n config.

to test if everything works fine, try a configuring ur CBWFQ bandwidth allocation to a solid sum of over 768K bandwidth under ur policy-map, see if you get any error.

policy-map test

class er

  priority 512

class af11

  bandwidth 256

class af21

  bandwidth 256

if your config is correct, when you assign this policy-map you MUST get an error.

show policy-map interface s0/0  to verify.

plz Rate if it helped,

Soroush.

[durale1789]

policy is bound to the interface