Re: OT: Catalys 6500 qos policing doubts

Juan,
the 6k line is a complicated beast. QoS is done at several places, and
it is an evolving architecture.

That said, the page your are referring seems to address PFC QoS, i.e.,
QoS implemented by the policy feature card at a global stage.
(As opposed to line card QoS)

This same page says:

   You can classify and apply QoS (for example, marking) and then, on
   another interface or network device, classify again based on the
   marked value and apply other QoS.

The "markdown DSCP" is (I'm interpreting the doc now, i.e., I'm not
talking from experience but from an educated interpretation) a rewrite
of the incoming DSCP based on policing. (That is, if you exceed some
limit, then transmit and remark).

The way I read the document, it means that the remarking of a given
internal DSCP is global. That is, if an exceed clause remarks AF31 into
AF43, it has to do so everywhere and you can not remark AF31 into AF41
elsewhere in the same chasis.

There is another interesting note:

   When you apply both ingress policing and egress policing to the same
   traffic, both the input policy and the output policy must either mark
   down traffic or drop traffic. PFC QoS does not support ingress
   markdown with egress drop or ingress drop with egress markdown.

which might make sense because PFC logic is applied once in the packet
flow, not twice, and somehow combining a drop with a remark is not
supported.

Hope this makes sense. Or else, causes someone with more info to speak up :)

Also, I've found lots of useful QoS information in cisco's medianet
design pages, at
http://www.cisco.com/en/US/docs/solutions/Enterprise/Video/Medianet_Ref_Gd/medianet_ref_gd.html
and
http://www.cisco.com/en/US/docs/solutions/Enterprise/WAN_and_MAN/QoS_SRND_40/QoSWAN_40.html

-Carlos

Juan @ 21/05/2012 18:37 -0300 dixit:
> Carlos, maybe i am misunderstanding the 15.0.1 IOS for 6500 config notes
> , but at:
> http://www.cisco.com/en/US/docs/switches/lan/catalyst6500/ios/15.0SY/configuration/guide/qos_class_mark_police.html
> you can see this note at "Overview of Policing":
>
> "For all policers, PFC QoS uses a configurable global table that maps
> the internal DSCP value to a marked-down DSCP value. When markdown
> occurs, PFC QoS gets the marked-down DSCP value from the table. You
> cannot specify marked-down DSCP values in individual policers."
>
> I am trying to do this:
> -mark some types of traffic entering Cat6500 layer 2 interfaces using
> different DSCP values
> -policing these same flows previously marked at a different output
> interface (layer 3 interface of the same 6500) , by matching them
> using dscp values "assigned" on the input. Then, i do not see clearly if
> the comment "You cannot specify marked-down DSCP values in individual
> policers." is a restriction for this type of config (i have done a SPAN
> of the output traffic, and i can not see the new dscp values... maybe
> this is normal because the dscp assigned at the input is only an
> "internal dscp" ??)
> Many thanks newly & Best Regards
>
>
> 2012/5/21 Carlos G Mendioroz <tron_at_huapi.ba.ar <mailto:tron_at_huapi.ba.ar>>
>
> Where did you read that ?
> AFAIK, you can mark on input and do congestion management on output
> even on a 6k.
> -Carlos
>
> JuanF @ 21/05/2012 15:40 -0300 dixit:
>
> Hi all,
>
> I have a 6509 Catalyst with:
> -Supervisor card 2T
> -Ios 15.0.SY1
>
> I need to do a QoS policing configuration on this hardware, with
> these
> conditions:
>
> -Mark Dscp value from some types of IP traffic on several layer2
> INPUT
> Interfaces of the 6509
> -Use these DSCPs marks to aply different QoS policings to these
> flows
>
> Is it possible to do all this by configuring only the 6500
> switch? (this is,
> without depending of matching the dscp value marked by a
> diferent device). I
> read you can not mark the Dscp on the 65000 and also manipulate
> the same
> traffic using these Dscps values locally asigned on the 65000.
>
> TIA and Best Regards
>
>
> Blogs and organic groups at http://www.ccie.net
> <http://www.ccie.net/>
>
> ___________________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/__list/CCIELab.html
> <http://www.groupstudy.com/list/CCIELab.html>
>
>
>
>
>
>
>
>
> --
> Carlos G Mendioroz <tron_at_huapi.ba.ar <mailto:tron_at_huapi.ba.ar>> LW7
> EQI Argentina
>
>
>
>
> --
> Juan Faure

-- 
Carlos G Mendioroz  <tron_at_huapi.ba.ar>  LW7 EQI  Argentina
Blogs and organic groups at http://www.ccie.net