From: Bob Sinclair (bsin@cox.net)
Date: Mon Nov 22 2004 - 15:36:41 GMT-3
Tim,
Perhaps you could create an input policy that matches on dlci 101 and
associates with a qos group. The create an output policy with two classes:
a) match qos group, and match dlci 102 and b) match qos group. and match
dlci 103.
Bob Sinclair
CCIE #10427, CCSI 30427, CISSP
www.netmasterclass.net
----- Original Message -----
From: "ccie2be" <ccie2be@nyc.rr.com>
To: "Bob Sinclair" <bsin@cox.net>; "Group Study" <ccielab@groupstudy.com>
Sent: Monday, November 22, 2004 11:58 AM
Subject: Re: MQC Matching on DLCI
> Bob,
>
> Yep, that last example was fairly straighforward. Now, I'm trying to
> figure
> out if this is possible and exactly how this would be configured.
>
> Suppose you have a hub f/r router with 3 spokes, 1, 2, and 3, all
> terminating on a p2m interface.
>
> Configure the hub router such that traffic from spoke 1 going to spoke 2
> has
> action 2 done and traffic from spoke 1 going to spoke 3 has action 3 done.
>
> My problem is that I can't think of a way to match on next-hop address.
> If
> there were such a matching criteria, then this would be easy. I could just
> configure this.
>
> class-map 1-to-2
> match fr-dlci 101
> match <next-hop of spoke-2>
>
> class-map 1-to-3
> match fr-dlci 101
> match <next-hop of spoke-3>
>
> policy-map Actions
> class 1-to-2
> set <Action-2>
> class 1-to-3
> set <Action-3>
>
>
> int s0
> encap fram
> service-policy out Actions
>
> I don't know of any way to code "match <next-hop of spoke X>", so I don't
> know if this can be done.
>
> Any thoughts?
>
> Tim
>
>
> ----- Original Message -----
> From: "Bob Sinclair" <bsin@cox.net>
> To: "ccie2be" <ccie2be@nyc.rr.com>; "Group Study" <ccielab@groupstudy.com>
> Sent: Monday, November 22, 2004 11:16 AM
> Subject: Re: MQC Matching on DLCI
>
>
>> Tim,
>>
>> Absolutely. The policy will set precedence to X for all traffic outbound
>> DLCI 102 on the hub, regardless of its source.
>>
>> Bob Sinclair
>> CCIE #10427, CCSI 30427, CISSP
>> www.netmasterclass.net
>>
>>
>> ----- Original Message -----
>> From: "ccie2be" <ccie2be@nyc.rr.com>
>> To: "Bob Sinclair" <bsin@cox.net>; "Group Study" <ccielab@groupstudy.com>
>> Sent: Monday, November 22, 2004 10:37 AM
>> Subject: Re: MQC Matching on DLCI
>>
>>
>> > Bob,
>> >
>> > Let's say R1 has 1 dlci, 102 , to R2 adn another one to R3, 103.
>> >
>> > Couldn't this be done?
>> >
>> > class-map DLCI-102
>> > match fr-dlci 102
>> >
>> > policy-map IP-PREC
>> > class DLCI-102
>> > set ip-prec X
>> >
>> > int s0
>> > encap fram
>> > fram map ip x.x.x.2 102 broadcast
>> > fram map ip x.x.x.3 103 broadcast
>> > service-policy output IP-PREC
>> >
>> > Now, traffic to R2 will have the ip prec set to X while the dlci to R3
>> > will
>> > just have the default precedence, don't ya think?
>> >
>> > Tim
>> > ----- Original Message -----
>> > From: "Bob Sinclair" <bsin@cox.net>
>> > To: "ccie2be" <ccie2be@nyc.rr.com>; "Group Study"
> <ccielab@groupstudy.com>
>> > Sent: Monday, November 22, 2004 10:18 AM
>> > Subject: Re: CAR vs Police
>> >
>> >
>> >> Tim,
>> >>
>> >> I too have trouble visualizing it! I went back and checked my lab
> notes,
>> >> and what I actually did was to set precedence on the hub for traffic
>> >> transiting between spokes. The policy was applied on input matching
>> >> on
> a
>> >> DLCI. I also set precedence using an input policy and matched that
>> >> precedence in an output policy on the same multipoint. I cannot think
> of
>> > a
>> >> way to match input DLCI in an outbound policy, as you can with an
>> >> input
>> >> interface match.
>> >>
>> >> Good catch!
>> >>
>> >> Bob Sinclair
>> >> CCIE #10427, CCSI 30427, CISSP
>> >> www.netmasterclass.net
>> >>
>> >>
>> >> ----- Original Message -----
>> >> From: "ccie2be" <ccie2be@nyc.rr.com>
>> >> To: "Bob Sinclair" <bsin@cox.net>; "Group Study"
> <ccielab@groupstudy.com>
>> >> Sent: Monday, November 22, 2004 10:04 AM
>> >> Subject: Re: CAR vs Police
>> >>
>> >>
>> >> > Hey Bob,
>> >> >
>> >> > All in all, this MQC stuff is turning to be very flexible and
> powerful.
>> >> >
>> >> > But, I can't visualize how to configure that matching inbound and
>> > setting
>> >> > outbound thing you mentioned. A service policy is one way either
>> > inbound
>> >> > or
>> >> > outbound. Any chance you can show me an example of what you're
> saying?
>> >> >
>> >> > Thanks, Tim
>> >> >
>> >> > ----- Original Message -----
>> >> > From: "Bob Sinclair" <bsin@cox.net>
>> >> > To: "ccie2be" <ccie2be@nyc.rr.com>; "Group Study"
>> > <ccielab@groupstudy.com>
>> >> > Sent: Monday, November 22, 2004 9:37 AM
>> >> > Subject: Re: CAR vs Police
>> >> >
>> >> >
>> >> >> Tim,
>> >> >>
>> >> >> I have had a chance to lab up matching on DLCIs, and it does behave
> as
>> >> >> you
>> >> >> indicate: applies to outbound traffic if the policy is applied
>> > outbound,
>> >> >> and vice versa. What caught my eye: you can match inbound DLCI
>> >> >> and
>> > set
>> >> >> outbound, even when both DLCIs are on a single multipoint. Another
>> >> >> illustration that the multipoint hub does a complete
>> >> >> decapsulation -
>> >> > routing
>> >> >> process - encapsulation.
>> >> >>
>> >> >> Bob Sinclair
>> >> >> CCIE #10427, CCSI 30427, CISSP
>> >> >> www.netmasterclass.net
>> >> >>
>> >> >>
>> >> >> ----- Original Message -----
>> >> >> From: "ccie2be" <ccie2be@nyc.rr.com>
>> >> >> To: "ccie2be" <ccie2be@nyc.rr.com>; "Bob Sinclair" <bsin@cox.net>;
>> > "Anas
>> >> >> Tarsha" <ra3i@yahoo.com>; "Group Study" <ccielab@groupstudy.com>
>> >> >> Sent: Monday, November 22, 2004 8:58 AM
>> >> >> Subject: Re: CAR vs Police
>> >> >>
>> >> >>
>> >> >> > Well, I think I found the answer to my questions regarding CAR vs
>> >> >> > Police
>> >> >> > using MQC.
>> >> >> >
>> >> >> > With the new features now available, there is NO functionality in
>> >> >> > CAR
>> >> > that
>> >> >> > isn't available with MQC.
>> >> >> >
>> >> >> > See this link:
>> >> >> >
>> >> >> >
>> >> >
>> >
> http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122newft/122t/122t13/ft3level.htm
>> >> >> >
>> >> >> > As this link shows, it's now possible to nest police commands
>> >> >> > with
>> > MQC.
>> >> >> >
>> >> >> > And, it's also possible to match traffic based on dlci, so
> therefore
>> > a
>> >> >> > policy can be applied to the traffic of one particular dlci.
>> >> >> >
>> >> >> >
>> >> >
>> >
> http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122newft/122t/122t13/ftpcdlci.htm
>> >> >> >
>> >> >> > I assume that when matching traffic based on dlci, whether the
> dlci
>> >> >> > matched
>> >> >> > is the incoming or outgoing depends on whether the service-policy
> is
>> >> >> > applied
>> >> >> > to outgoing or incoming traffic, but I can't confirm that until I
>> >> >> > get
>> >> > some
>> >> >> > rack time.
>> >> >> >
>> >> >> > Any thought?
>> >> >> >
>> >> >> > Tim
>> >> >> > ----- Original Message -----
>> >> >> > From: "ccie2be" <ccie2be@nyc.rr.com>
>> >> >> > To: "Bob Sinclair" <bsin@cox.net>; "Anas Tarsha"
>> >> >> > <ra3i@yahoo.com>;
>> >> > "Group
>> >> >> > Study" <ccielab@groupstudy.com>
>> >> >> > Sent: Sunday, November 21, 2004 5:35 PM
>> >> >> > Subject: Re: CAR vs Police
>> >> >> >
>> >> >> >
>> >> >> >> Hi Bob,
>> >> >> >>
>> >> >> >> As I going over all the old GS posts on policing, I came this
>> >> >> >> one
>> >> >> > comparing
>> >> >> >> CAR and Police.
>> >> >> >>
>> >> >> >> According to the DQoS by Odem, one of the differences is that
> with
>> > CAR
>> >> >> >> you
>> >> >> >> can configure nested rate-limit commands but not with MQC.
>> >> >> >> Since
>> >> > there's
>> >> >> >> been alot of new features added to MQC, I wonder if that still
>> >> >> >> holds
>> >> >> >> true.
>> >> >> >>
>> >> >> >> For example, according to Odem, you couldn't enable MQC on a per
>> > dlci
>> >> >> > basis,
>> >> >> >> but since now you can match on dlci, you can.
>> >> >> >>
>> >> >> >> So, I wonder if a config like this would fly
>> >> >> >>
>> >> >> >> policy-map VOICE
>> >> >> >> class VOICE
>> >> >> >> police 128000
>> >> >> >>
>> >> >> >> policy-map ALL-TRAFFIC
>> >> >> >> class class-default
>> >> >> >> police 256000
>> >> >> >> service-policy VOICE
>> >> >> >>
>> >> >> >> Any thoughts?
>> >> >> >>
>> >> >> >> Thanks, Tim
>> >> >> >>
>> >> >> >> ----- Original Message -----
>> >> >> >> From: "Bob Sinclair" <bsin@cox.net>
>> >> >> >> To: "Anas Tarsha" <ra3i@yahoo.com>;"Group Study"
>> >> > <ccielab@groupstudy.com>
>> >> >> >> Sent: Saturday, December 06, 2003 9:51 PM
>> >> >> >> Subject: Re: CAR vs Police
>> >> >> >>
>> >> >> >>
>> >> >> >> > Anas,
>> >> >> >> >
>> >> >> >> > I would agree that Cisco could do a better job of explaining
> the
>> >> >> > policing
>> >> >> >> > mechanism in the MQC, but I think I would disagree that it
>> >> >> >> > permits
>> >> >> >> buffering
>> >> >> >> > during congestion. There are Bc and Be parameters, but
> according
>> > to
>> >> >> >> > the
>> >> >> >> > documentation this does not buffer packets to shape the
> traffic.
>> >> >> >> > The
>> >> >> > best
>> >> >> >> > explanation I have found is at the link below, which says that
>> >> >> >> > the
>> >
>> >> >> > policer
>> >> >> >> > does not buffer, but "drops packets less aggressively" during
>> >> >> > congestion.
>> >> >> >> > Could you check it out and see what you think of it?
>> >> >> >> >
>> >> >> >> >
>> >> >> >>
>> >> >> >
>> >> >
>> >
> http://www.cisco.com/en/US/partner/tech/tk543/tk545/technologies_q_and_a_item09186a00800cdfab.shtml#Q24
>> >> >> >> >
>> >> >> >> > Thanks!
>> >> >> >> >
>> >> >> >> > -Bob Sinclair
>> >> >> >> > CCIE #10427, CISSP, MCSE
>> >> >> >> >
>> >> >> >> >
>> >> >> >> > bsinclair@netmasterclass.net
>> >> >> >> > ----- Original Message -----
>> >> >> >> > From: "Anas Tarsha" <ra3i@yahoo.com>
>> >> >> >> > To: "Bob Sinclair" <bsin@cox.net>; "ccie2be"
>> >> >> >> > <ccie2be@nyc.rr.com>;
>> >> >> > "Group
>> >> >> >> > Study" <ccielab@groupstudy.com>
>> >> >> >> > Sent: Saturday, December 06, 2003 9:07 PM
>> >> >> >> > Subject: Re: CAR vs Police
>> >> >> >> >
>> >> >> >> >
>> >> >> >> > > Also theoretically there is a major difference between
>> >> >> >> > > CAR and policing, a difference which Cisco hardly
>> >> >> >> > > explains it well in my opinion. CAR is a rate-limit
>> >> >> >> > > mechanism to limit the input or output transmission
>> >> >> >> > > rate on an interface or subinterface based on a
>> >> >> >> > > configured value. All the exceeding traffic is dropped
>> >> >> >> > > in case the exceeding action is dropping. Policing is
>> >> >> >> > > more like a shaping mechanism. As the name implies,
>> >> >> >> > > shaping does not drop packets in case of congestion,
>> >> >> >> > > it buffers them. You will see delay but no data loss.
>> >> >> >> > > So this is the major difference, CAR does not buffer.
>> >> >> >> > >
>> >> >> >> > > Anas
>> >> >> >>
>> >> >> >>
>> > _______________________________________________________________________
>> >> >> >> Subscription information may be found at:
>> >> >> >> http://www.groupstudy.com/list/CCIELab.html
>> >> >> >
>> >> >> >
>> > _______________________________________________________________________
>> >> >> > Subscription information may be found at:
>> >> >> > http://www.groupstudy.com/list/CCIELab.html
>> >> >
>> >> >
> _______________________________________________________________________
>> >> > Subscription information may be found at:
>> >> > http://www.groupstudy.com/list/CCIELab.html
>> >
>> > _______________________________________________________________________
>> > Subscription information may be found at:
>> > http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Thu Dec 02 2004 - 06:57:48 GMT-3