Re: WFQ: Rounded packet size

From: Bit Gossip <bit.gossip_at_chello.nl>
Date: Mon, 26 Apr 2010 22:09:59 +0200

This is indeed confirmed by the following article:

http://www.cisco.com/en/US/partner/tech/tk39/tk824/technologies_tech_note09186a00800c9699.shtml#topic4

On Mon, 2010-04-26 at 12:57 -0700, Petr Lapukhov wrote:
> Using packet size for SN increment is crucial for fair-queueing. This
> way it allows for bandwidth allocation based on the actual traffic
> rate. Clearly, if you add just the weight, you assume all flows have
> equal packet sizes.
>
> HTH
>
> 2010/4/26 Bit Gossip <bit.gossip_at_chello.nl>:
> > Petr and all,
> > the formula for active flow from the IEWB below is:
> > newSN = previousSN + W * PacketSize
> >
> > This is very different from the formula in "CCIE Practical Studies
> > Volume II" chapter "Chapter 6:
> > newSN = previousSN + W
> >
> > They are profoundly different because the latter doesn't account for
> > packet size.
> >
> > So which is the correct one?
> > Thanks,
> > bit.
> >
> >
> > On Mon, 2008-12-15 at 15:56 +0300, Petr Lapukhov wrote:
> >> Hi,
> >>
> >>
> >> Most likely, the book refers to the "round number" which is the size
> >> of the last packet sent by the scheduler. This value is used to
> >> initialize every new flow queue byte counter. You can read a detailed
> >> description of WFQ algorithm (explaning how this round number is used
> >> btw) here:
> >>
> >>
> >> http://www.internetworkexpert.com/downloads/IEWB-RS-VOL-I-V5.Section.10.QoS.teaser.pdf
> >>
> >>
> >> The second task in dedicated to WFQ. Understanding the internal
> >> working of WFQ is very important since the same scheduling procedure
> >> is used by CBWFQ.
> >>
> >>
> >> HTH,
> >>
> >>
> >> --
> >> Petr Lapukhov, CCIE #16379 (R&S/Security/SP/Voice)
> >> petr_at_internetworkexpert.com
> >>
> >> Internetwork Expert, Inc.
> >> http://www.InternetworkExpert.com
> >> Toll Free: 877-224-8987
> >> Outside US: 775-826-4344
> >>
> >>
> >>
> >> 2008/12/14 Bit Gossip <bit.gossip_at_chello.nl>
> >> Experts,
> >> any idea on how these value are calculated for the WFQ min-max
> >> algorithm?
> >> In "CCIE Practical Studies Volume II" chapter "Chapter 6. QoS
> >> b Rate
> >> Limiting and Queuing Traffic" there is an example where 60 and
> >> 700 are
> >> used. But how are these value determined?
> >> "During the WFQ process, packet size is noted when new packets
> >> arrive.
> >> If the IP flow that they belong to is new, a rounded packet
> >> size is also
> >> used."
> >> Tnx,
> >> bit
> >>
> >>
> >> Blogs and organic groups at http://www.ccie.net
> >>
> >> _______________________________________________________________________
> >> Subscription information may be found at:
> >> http://www.groupstudy.com/list/CCIELab.html
> >
> >
> > Blogs and organic groups at http://www.ccie.net
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html

Blogs and organic groups at http://www.ccie.net
Received on Mon Apr 26 2010 - 22:09:59 ART

This archive was generated by hypermail 2.2.0 : Sat May 01 2010 - 09:49:57 ART