[ovs-discuss] QoS in OVS compared to tc

[Ben Pfaff]

To clarify Ethan's response just a little: OVS uses the same underlying
mechanism as "tc", but it does not actually invoke the "tc" binary.

On Tue, Aug 16, 2011 at 05:35:39PM -0700, Syed Akbar Mehdi wrote:
> As a follow-up question to this, does OVS use tc at the backend to configure
> queues? I ask this because lets say I want to configure queues using tc and
> then somehow have them reflected in the OVS configuration database, so that
> later I can map flows to these queues (e.g. using Openflow). Is this
> possible?
> 
> Thanks,
> Akbar
> 
> On Tue, Aug 16, 2011 at 5:09 PM, Syed Akbar Mehdi <
> akbar.mehdi at seecs.nust.edu.pk> wrote:
> 
> > Hi,
> >
> > The description of the QoS tables in the Open vSwitch configuration
> > database (http://openvswitch.org/ovs-vswitchd.conf.db.5.pdf manual on page
> > 21-22) refers to the documentation of the Linux hierarchy token bucket
> > classifier at http://luxik.cdi.cz/~devik/qos/htb/manual/userg.htm.
> >
> > However if I configure the Linux-htb classifier using tc, I get the option
> > to configure qdiscs and classes in a "hierarchy" which allows fine-grained
> > tuning of QoS policies e.g. excess-bandwidth sharing (or not sharing)
> > between different classes. The OVS documentation gives the impression that
> > this hierarchical configuration is not possible when configuring queues
> > using OVS (e.g. using ovs-vsctl). It appears that I can only configure
> > queues in a "flat" configuration where all queues are at the same level. Is
> > this true? In case it is then what is the policy for excess bandwidth
> > sharing between different queues.
> >
> > Regards,
> > Akbar
> >
> 
> 
> 
> -- 
> Regards,
> Syed Akbar Mehdi,
> School of EECS (SEECS),
> National University of Sciences and Technology (NUST),
> Pakistan.

> _______________________________________________
> discuss mailing list
> discuss at openvswitch.org
> http://openvswitch.org/mailman/listinfo/discuss