[c-nsp] CBWFQ-LLQ on PPPoE Virtual Templates

[Dean Smith]

I think our lab has just fallen over something very similar.

Something goes wrong with QoS with lots of PPPoVPDN sessions on
12.2(31)SB11. Works perfectly with a few on a test rig. Load it up and it
starts to fail.

We've tested (just this feature) on 12.2(33)SRC and it works again. (This is
on 7201)

Dean Smith

-----Original Message-----
From: cisco-nsp-bounces at puck.nether.net
[mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of David Granzer
Sent: 10 April 2008 15:35
To: Gregory Boehnlein
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] CBWFQ-LLQ on PPPoE Virtual Templates

On 4/10/08, Gregory Boehnlein <damin at nacs.net> wrote:
> > > Hello,
>  > >     We have a 7206 running (C7200-IK9S-M), Version 12.3(20), RELEASE
>  > > SOFTWARE (fc2). We are trying to get LLQ implemented on Virtual
>  > > Template interfaces for our PPPoE DSL users:
>  > >
>  > > vpdn-group akrnaa01rr
>  > >  description SBC Akron VPDN Group
>  > >  accept-dialin
>  > >   protocol l2tp
>  > >   virtual-template 1
>  > >  terminate-from hostname akrnaa01rr.oh.AADS
>  > >  local name xxxxxx
>  > >  lcp renegotiation always
>  > >  l2tp tunnel password 7 xxxxxxxxxxx
>  >
>  > QoS for L2TP users is not supported in your release, you need 12.2SB,
>  > take a look at the "Per-Session QoS" feature at
>  > http://www.cisco.com/en/US/docs/ios/12_2sb/feature/guide/sbbbrs1b.html
>
>
> Excellent. Thank you very much. What does the "SB" designation stand for?

There are two SB releases, 122-28.SB and 122-31.SB. I'm not surre what
exactly the difference is between the two, but we have curious
experience with 122-31.SB releases with Per-Session QoS (traffic
shaping) enabled. If there a lot of PPPoE sessions (more than 3000) on
the 7301 the  throughput per session goes notably down. We didn't see
this behavior with 122-28.SB.

David

>  > > interface Virtual-Template1
>  > >  mtu 1492
>  > >  ip unnumbered Loopback1
>  > >  rate-limit output access-group 102 8000 1500 2000 conform-action
>  > > transmit exceed-action drop
>  > >  no ip route-cache cef
>  > >  no ip route-cache
>  >
>  > Disabling CEF is always a bad idea, but
>  >
>  > >  compress stac
>  >
>  > causes all your traffic to be process switched anyway. You don't want
>  > to use compression on anything faster than ISDN/64k, it doesn't scale..
>
>
> That looks to be a holdover from the old ISDN template that this was
>  converted from. I'll consider removing STAC compression from the mix.
>
>
>
>  _______________________________________________
>  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/
>
_______________________________________________
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/