[c-nsp] MPLS to IP QoS on 7600 platform

Cydon Satyr cydonsatyr at gmail.com
Thu Jan 30 07:47:16 EST 2014


Default is short-pipe, maybe I wasn't clear enough. The problem is we have
pipe mode with explicit-null signaling in our core. But I wasn't able to
find how to classify outgoing MPLS-to-IP traffic based on this EXP.
In L3VPN with two label stack, mpls propagate-cos copies EXP to DSCP, but
if there is only one label (explicit-null) then mentioned command doesn't
work.

Scenario: Ip packet with cos0/cs0 enters PE router. PE router puts EXP3 and
forwards. Label egress PE gets explicit null label with EXP3. Router
removes EXP3, and sends IP packet with cs0.
Because the above command doesn't work, this means that egress PE router
can't rewrite packet's dscp with exp. This means that packet has to contain
valid dscp PRIOR to entering ingress PE router, as otherwise there's no way
to change dscp. On 7600 at least (I think SIP600 could use qos-groups but
that's another topic. Problem are LAN cards).

Could anyone agree/disagree on this?

Kind Regards




On Thu, Jan 30, 2014 at 1:23 PM, Adam Vitkovsky <adam.vitkovsky at swan.sk>wrote:

> Hello,
>
> That doesn't seem to be correct I believe all mpls capable cisco boxes
> defaults to "Short Pipe" tunneling mode
>
> http://www.cisco.com/en/US/docs/routers/7600/ios/15S/configuration/guide/mpl
> sqos.html#wp1487531
> Bottom of the table confirms:
> MPLS to IP edge QoS  - Preserve the exposed IP DSCP
>
>
> adam
> -----Original Message-----
> From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of
> Cydon Satyr
> Sent: Wednesday, January 29, 2014 9:35 PM
> To: cisco-nsp at puck.nether.net
> Subject: [c-nsp] MPLS to IP QoS on 7600 platform
>
> Hello,
>
>
> I hope you can make this a bit clearer.
>
>
> Reading up on Cisco documentation for 7600 router, there are two outcomes
> of
> popping labels and forwarding an IP packet in regards to what value is
> trusted for internal DSCP.
>
>
> a) If a label is aggregate then ultimately IP packet's DSCP value is used
> to
> make internal DSCP which is then used to write customer 802.1q tag (COS).
>
>
> b) If it is not aggregate, then exp is trusted, in which case exp is used
> for internal DSCP, and finally customer COS.
>
>
> Why different behavior? By default directly connected vrf routes will be
> treated with option a) and other customer routes (per prefix vpn label)
> will
> get COS based on EXP?
>
>
> Also, I think the only way to rewrite DSCP is with (hidden) command mpls
> propagate-cos on PE router's CE-facing L3 interface.
>
>
> BUT what if we want to simply copy value from EXP in MPLS packet to DSCP
> packet for non-L3VPN service? In other words a router will get 'explicit
> null' label, but the above command will not copy value from EXP->Internal
> DSCP-> DSCP. Also, there are no qos-groups on 7600.
>
>
> But that means that neither Ingress PE nor Egress PE router cannot change
> DSCP for GRT routes that go over MPLS - change has to be done before it
> enters PE router, correct?
>
> Sorry if a bit too long. Was trying to be as concise as possible.
>
>
>
> Best Regards
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
>


More information about the cisco-nsp mailing list