Re: QOS help

Hi.
You can easily mitigate this by using HQF.
http://www.cisco.com/en/US/docs/ios/qos/configuration/guide/qos_frhqf_support.html

Just employ the correct hierarchy.

Here is a simple example:
Say you have 3 classes in your QoS domain: C1, C2, and C3.
All of them need to get the guarantees needed.
This invloves minimum capacity reservation (bandwidth), latency control
(queue-limit). For example:

C1
bandwidth percent 20
queue-limit 25 ms
C2
bandwidth percent 5
queue-limit 5 ms
C3
bandwidth percent 75
queue-limit 40 ms
Now, you need to do this per site, at the hub site.

Let's say your three sites have IP ranges as follows:
SITE1: 11.1.1.0/24
SITE2: 22.2.2.0/24
SITE3: 33.3.3.0/24
So you create an ACL for each site:
SITE1-ACL permit any 11.1.1.0 0.0.0.255
SITE2-ACL permit any 22.2.2.0 0.0.0.255
SITE3-ACL permit any 33.3.3.0 0.0.0.255

And classes for each site, based on the site ACL:
SITE-CLASS-1
 match access-group name SITE1-ACL
 SITE-CLASS-2
 match access-group name SITE2-ACL
 SITE-CLASS-3
 match access-group name SITE3-ACL

Then you shape traffic to each site, and apply your domain policy to the
resulting shaper:

policy-map DOMAIN-POLICY
  class C1
  bandwidth percent 20
  queue-limit 25 ms
 class C2
  bandwidth percent 5
  queue-limit 5 ms
 class C3
  bandwidth percent 75
  queue-limit 40 ms

policy-map SOME-HQF-POLICY
  class SITE-CLASS-1
   shape average 8m
   service-policy DOMAIN-POLICY
  class SITE-CLASS-2
    shape average 100m
   service-policy DOMAIN-POLICY
   class SITE-CLASS-3
    shape average 100m
   service-policy DOMAIN-POLICY

Finally, apply that HQF to the output interface at the hub (prey in the mean
time that your HW support HQF :) meaning it's either SW router (x8xx, x9xx
or 7200) or it's an ASR1K)
interface gigabit0/0/0
service-policy output SOME-HQF-POLICY

**Note: having said that, normally this is all done by the ISP at the egress
of the spoke (remote) site. E.g. You send your traffic from the hub site
traveling at 1G, ISP transports the traffic over its 10G+ core, and it
arrives on the PE which has the slow last mile connected (8M or 100M). That
PE is the one that normally has your policy maps implemented. ISPs offer
this for as low as $20/month, and its considered standard these days.

On Mon, Sep 19, 2011 at 11:30 PM, shady darwish <engshad.shady_at_gmail.com>wrote:

> i have a router at HQ with 400 mbps port speed connected as p-to-p to
> services provider and this port mapped to 3 remote sites :-
>
> A 8 mbps oprt speed
> B 100 mbps port speed
> C 100 mbps port speed
>
> Note that all remote sites are in different subnets
>
> the problem is i have traffic over low from HQ to site C and i need to have
> nested QOS guarantee priority for VOIP all the time and in same restrict
> the
> BW for each remote site and priority for the corresponding voice services
> BW
> without drop any packets at any time time
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
>

-- 
Pavel Bykov
Blogs and organic groups at http://www.ccie.net