[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: CBQ scheduling




Hi, Urs


Urs Hengartner wrote:
> 
> I'm observing behavior for CBQ scheduling that I cannot explain.

A similar thing has happened to me.


> Bandwidth distribution works as expected unless I configure the two
> child classes to be non-bounded. In that case, both flows get about
> the same bandwidth. Is there a logical explanation for this behavior?

When I define the classes as "bounded", they get the correct bandwith.
The problem is that they can't borrow any unused bandwidth :(

Some weeks ago, somebody on this answered a question of mine on this
same subject. I quote his answer with double ">":

>>In regards to CBQ class differentiation:
>>
>>* the first differentiator between classes is the priority; the lower >>the
>>value, the more important the class is; 
>>* the second is b/width: if you give wire rate speed eg 10 Mbps on a
>>10Mbps line, then it will starve the least important class.
>>* Weight and allot could also be used to differentiate: In the case >>where
>>you have given equal priority to classes then each class's quantum is
>>defined by the product: 
>>itsweight*itsallot*num_classes/Sum(allclassweights) 
>>"it" meaning the class.

You have specified the same priority and the same weight for the two
classes. Maybe if you stated weights 9 and 1 for the two classes?

PLEASE, if you try this out and it works OK, let me know, since I'm just
working on this same feature at the time.

Regards,

	Mariano Korman