Re: DSCP AF class

From: Jay Pal <jay.b.pal_at_gmail.com>
Date: Wed, 22 Apr 2009 11:17:57 +0800

Bryan and Ravi,
Thanks for the explanation. much more comfortable now. this is why I love
this groupstudy forum.

One more question before i get back to my lab work. What are other things
that can influence a particular queue (LLQ, bandwidth command per class,
changing the drop value on RED/WRED per class/DSCP, policier)?

thanks
Jay

On Wed, Apr 22, 2009 at 10:56 AM, Ravi Singh <way2ccie_at_googlemail.com>wrote:

> Hi Jay,
>
> AF4 does not "have" a better queuing treatment than AF1. It "should" ,
> as per the defined AF PHB.As WorkerBee said "following the standards
> help to improve the QoS treatment of properly marked DSCP packets ".So
> in a live environment, you may want to prefer AF4 over AF1 markings
> but If you are working in a lab with two routers and you have incoming
> packets marked as DSCP AF4x and AF1x, there is no default preference
> of the queuing treatment they will get, since they are both classified
> into different classes and packets in different classes , for example
> AF4 and AF1, are processed independently of each other. It's when you
> explicitly configure a queuing mechanism, as in Bryan's example, that
> you could affect the preference of one class over another.
>
> HTH,
> Ravi
>
> On Wed, Apr 22, 2009 at 3:40 AM, Jay Pal <jay.b.pal_at_gmail.com> wrote:
> > Hi Ravi, Gaurav, Bryan and WorkerBee,
> > thanks for the inputs/explaination.
> > This is where it gets confusing to me. The Qos book (mentioned earlier in
> > the post by Ravi) that 'there is no inherent advantage to being in class
> 4
> > versus class 1'. And in another post Ravi mentions that
> > 'For example,
> > packets in class AF4 should be given better queuing treatment than
> > packets in AF1.In other words, when you classify and mark packets and
> > assign them the AF PHB, a packet with a PHB of AF4 is not bound to
> > receive better queuing treatment than the AF1 packet.'
> > So I am confused as to why AF4 has better queue treatment than AF1 in a
> DSCP
> > aware network. I can understand that if the network is not DSCP aware
> then
> > it would give better treatment for AF4 than AF1 (e.g. when you just have
> > WFQ on the interface). I have read few materials on the web and an RFC
> but
> > it still not quite clear.
> >
> > Bryan, Just to put another requirement on this scenario, lets just say
> that
> > you already have LLQ with voice running on EF. You could probably use CS
> 4
> > and CS1 to achieve this but i am trying to see if there is any other way
> of
> > doing it by using AF41 and AF11.
> >
> > thanks
> > Jay
> >
> > On Wed, Apr 22, 2009 at 10:01 AM, Ravi Singh <way2ccie_at_googlemail.com>
> > wrote:
> >>
> >> Exactly, That's why I said if you have a network domain totally under
> >> your control you may fiddle with the markings. Forgot to add that
> >> obviously such non-standard markings will not be of much use when sent
> >> outside the domain to other DSCP compliant networks. Thanks for adding
> >> this up WorkerBee.
> >>
> >> Regards,
> >> Ravi
> >>
> >> On Wed, Apr 22, 2009 at 2:42 AM, WorkerBee <ciscobee_at_gmail.com> wrote:
> >> > Ravi/Jay,
> >> >
> >> > AF classification is meant to standardize QoS service offering. It is
> >> > well understood that EF class is known to every engineer that it meant
> >> > LLQ/Priority while Best Effort is DSCP 0 marking.
> >> >
> >> > Of course you can choose to reverse the logic and implement the other
> >> > way round but you won't be integrating very well with other DSCP
> >> > compliant networks.
> >> >
> >> > That's why, you don't have the luxury of control the entire domain
> >> > but following the standards help to improve the QoS treatment of
> >> > properly marked DSCP packets.
> >> >
> >> > Engineering is fun. :)
> >> >
> >> > On Wed, Apr 22, 2009 at 9:27 AM, Ravi Singh <way2ccie_at_googlemail.com>
> >> > wrote:
> >> >> Hi Jay,
> >> >>
> >> >> As per my understanding, while inherently there is no advantage of a
> >> >> packet in one class over a packet in another, the AF PHB states that
> >> >> packets with a higher value of the AF class should be given queuing
> >> >> preference over packets in a lower value AF class . For example,
> >> >> packets in class AF4 should be given better queuing treatment than
> >> >> packets in AF1.In other words, when you classify and mark packets and
> >> >> assign them the AF PHB, a packet with a PHB of AF4 is not bound to
> >> >> receive better queuing treatment than the AF1 packet. It is totally
> >> >> upto you on what queuing treatment to give packets in different AF
> >> >> classes . If you have a network domain totally under your control,
> you
> >> >> may wish to provide packets in the AF1 class to be given the best
> >> >> treatment,whatsoever.
> >> >>
> >> >> And while I am writing this, Bryan has already responded on how you
> >> >> could achieve this configuration. Thanks Bryan.
> >> >>
> >> >> Regards,
> >> >> Ravi
> >> >>
> >> >>
> >> >> On Tue, Apr 21, 2009 at 11:40 PM, Jay Pal <jay.b.pal_at_gmail.com>
> wrote:
> >> >>> Hi Ravi,
> >> >>> In this case I have the same understanding then 'all AF class have
> the
> >> >>> same
> >> >>> priority'.
> >> >>>
> >> >>> Is there anyway to make AF4 with higher priority than AF1?
> >> >>>
> >> >>> thanks
> >> >>> Jay
> >> >>>
> >> >>> On Tue, Apr 21, 2009 at 11:32 PM, Ravi Singh <
> way2ccie_at_googlemail.com>
> >> >>> wrote:
> >> >>>>
> >> >>>> Hi Jay,
> >> >>>>
> >> >>>> Below is a paste of some relevant lines from the Cisco QoS Exam
> Cert
> >> >>>> Guide which will make this clear
> >> >>>>
> >> >>>> "An individual PHB describes what happens in a single hop, most
> >> >>>> typically a router. In the case of AF,
> >> >>>> each PHB contains two separate QoS function, typically performed by
> >> >>>> two different QoS tools. The
> >> >>>> first function is queuing. Each router classifies the packets into
> >> >>>> four different classes, and packets
> >> >>>> from each class are placed in a separate queue.
> >> >>>>
> >> >>>> The AF PHB defines Congestion Avoidance as the second behavior that
> >> >>>> comprises the AF PHB.
> >> >>>>
> >> >>>> Unlike the CS PHB, AF does not follow the bigger-is- better
> >> >>>> logic for the AF DSCPs. First, AF11, AF12, and so on are names for
> >> >>>> DSCP values, not the binary
> >> >>>> of decimal equivalent. Given the names, at least you can
> >> >>>> think of the first digit after the AF to be the queuing
> >> >>>> classification for example, all AF4x code
> >> >>>> points are in the same class for queuing. No specific queuing
> >> >>>> parameters are implied for any of these
> >> >>>> classes, so there is no inherent advantage to being in class 4
> versus
> >> >>>> class 1.
> >> >>>>
> >> >>>> Similarly, the second numeric digit in the AF DSCP names imply the
> >> >>>> drop preference with
> >> >>>> 3 meaning highest likelihood of being dropped, and 1 meaning the
> >> >>>> least
> >> >>>> likelihood. In other words,
> >> >>>> inside a single class, an AFx3 DSCP would mean that these packets
> >> >>>> would be dropped more quickly
> >> >>>> (more aggressively) than AFx2, which would be dropped more
> >> >>>> aggressively than AFx1 packets. In
> >> >>>> the actual DSCP names, a bigger number for the second numeric digit
> >> >>>> actually implies a less desirable
> >> >>>> QoS behavior.
> >> >>>>
> >> >>>> HTH,
> >> >>>> Ravi
> >> >>>>
> >> >>>>
> >> >>>> On Tue, Apr 21, 2009 at 2:23 PM, Jay Pal <jay.b.pal_at_gmail.com>
> wrote:
> >> >>>> > ok. I am clear on the drop probability. what about the first
> >> >>>> > numerical
> >> >>>> > value. Does AF4 have preference over AF1?
> >> >>>> >
> >> >>>> > On Tue, Apr 21, 2009 at 9:19 PM, GAURAV MADAN
> >> >>>> > <gauravmadan1177_at_gmail.com>wrote:
> >> >>>> >
> >> >>>> >> correct !
> >> >>>> >>
> >> >>>> >> first integer is "class selector" and second is " drop
> precedence"
> >> >>>> >>
> >> >>>> >> for drop prec : 1,2,3 .... 1 means low 2 means medium and 3
> means
> >> >>>> >> high
> >> >>>> >>
> >> >>>> >> HTH
> >> >>>> >> Gaurav Madan
> >> >>>> >>
> >> >>>> >> On Tue, Apr 21, 2009 at 6:43 PM, Jay Pal <jay.b.pal_at_gmail.com>
> >> >>>> >> wrote:
> >> >>>> >>
> >> >>>> >>> Actually, drop preference is the second numeric value in a
> >> >>>> >>> particular
> >> >>>> >>> class
> >> >>>> >>> like AF11 (low drop), AF12 (medium drop) and AF13 (high drop).
> >> >>>> >>>
> >> >>>> >>> I was talking more on AF class (AF1, AF2, AF3 and AF4).
> >> >>>> >>>
> >> >>>> >>> On Tue, Apr 21, 2009 at 9:07 PM, Divin Mathew John
> >> >>>> >>> <divinjohn_at_gmail.com
> >> >>>> >>> >wrote:
> >> >>>> >>>
> >> >>>> >>> > AF1, AF2 etc... 1,2 is the drop preference. like AF1 will be
> >> >>>> >>> > dropped
> >> >>>> >>> > less than AF2 and all
> >> >>>> >>> > Thanking You
> >> >>>> >>> >
> >> >>>> >>> > Yours Sincerely
> >> >>>> >>> >
> >> >>>> >>> > Divin Mathew John
> >> >>>> >>> > divinjohn_at_gmail.com
> >> >>>> >>> > divin_at_dide3d.com
> >> >>>> >>> > http://www.dide3d.com
> >> >>>> >>> > +91 9945430983
> >> >>>> >>> > +91 9846697191
> >> >>>> >>> > +974 5008916
> >> >>>> >>> > PGP PUBLIC KEY BLOCK @
> >> >>>> >>> > http://www.dide3d.com/divin_Public_PGP_key.txt
> >> >>>> >>> > Sent from Bangalore, KA, India
> >> >>>> >>> >
> >> >>>> >>> >
> >> >>>> >>> > On Tue, Apr 21, 2009 at 6:31 PM, Jay Pal <
> jay.b.pal_at_gmail.com>
> >> >>>> >>> > wrote:
> >> >>>> >>> > > Hi all,
> >> >>>> >>> > >
> >> >>>> >>> > > Just to make sure my understanding is correct on DSCP AF
> >> >>>> >>> > > class.
> >> >>>> >>> > > AF1,
> >> >>>> >>> AF2,
> >> >>>> >>> > > AF3 and AF4 have the same priority when using DSCP and
> there
> >> >>>> >>> > > is
> >> >>>> >>> > > no
> >> >>>> >>> > > advantage of using AF4 to AF1 or AF2.
> >> >>>> >>> > >
> >> >>>> >>> > > If this is correct is there a way to make one of the AF
> class
> >> >>>> >>> > > with
> >> >>>> >>> more
> >> >>>> >>> > > priority than other?
> >> >>>> >>> > >
> >> >>>> >>> > > thanks
> >> >>>> >>> > > Jay
> >> >>>> >>> > >
> >> >>>> >>> > >
> >> >>>> >>> > > 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
> >> >>>> >
> >> >>>> >
> >> >>>> >
> _______________________________________________________________________
> >> >>>> > 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 Wed Apr 22 2009 - 11:17:57 ART

This archive was generated by hypermail 2.2.0 : Mon May 04 2009 - 07:39:12 ART