CBWFQ / Metro E Configuration Question

[Sam Oesterling]

Hello all -

I have a 50 Mbps metro ethernet connection between our main office, and our collocation site, where we store web servers, DR equipment and VPN access gateways.  I have two Cisco 3845 ISR's connected to the metro E circuit.  The interfaces on each router are configured as 100/full as requested by my ISP.  We are connected via ethernet to a fiber media converter.

As I understand, CBFWQ will not kick in until congestion occurs on an interface.  I also understand that the bandwidth command on an interface is to provide bandwidth related information to upper level protocols (like EIGRP, etc).

My question is that since the interface where I have CBWFQ configured on is at 100 Mbps, but my circuit is at 50Mbps, how can I get my routers to kick CBWFQ in when traffic demand exceeds 50Mbps+?  Does the bandwidth command on the interface control that as well?

Thanks for any suggestions.

Brantley

[Jerry Ye · Accepted solution]

You need to do HQoS. Here is a sample config:

policy-map PARENT

class class-default

  shape average 50000000

  service-policy CHILD

policy-map CHILD

class CLASS1

class CLASS2

... SNIP...

HTH,

jerry

[Jon Marshall · Accepted solution]

Brantley

So you want to apply CBWFQ to the 50Mbps of traffic ? If so then you need to use hierarchical QOS ie. you first apply a shaper to the interface to shape the traffic to 50Mbps. Then within that shaped traffic you can apply child policies which apply CBWFQ to the shaped traffic. See this link for configuration details -

http://www.cisco.com/en/US/customer/docs/ios/qos/configuration/guide/qos_frhqf_support.html

Jon

[Sam Oesterling]

Thanks guys!  This was just want I was looking for!

[Joseph W. Doherty]

Disclaimer

The     Author of this posting offers the information contained within this     posting without consideration and with the reader's understanding  that    there's no implied or expressed suitability or fitness for any   purpose.   Information provided is for informational purposes only and   should not   be construed as rendering professional advice of any kind.   Usage of  this  posting's information is solely at reader's own risk.

Liability Disclaimer

In     no event shall Author be liable for any damages whatsoever   (including,   without limitation, damages for loss of use, data or   profit) arising  out  of the use or inability to use the posting's   information even if  Author  has been advised of the possibility of  such  damage.

Posting

One possible "gotcha", if your provider is providing you 50 Mbps, this often means you get 50 Mbps of wire bandwidth.  Unfortunately, I don't believe all IOS shapers account for L2 overhead.  When they don't, if you need to increase the accuracy of your shaping, you need to reduce the shaper's CIR to allow for L2.  Also unfortunately, L2 overhead varies by packet/frame size, so for super accuracy you need to allow for worst case.

For run-of-the-mill traffic shaping, and assuming you're not running, sustained, 100% logical link capacity, setting the shaper to nominal CIR is fine.

[Sam Oesterling]

I did not think of L2 overhead, but are not running at 100% all of the time.  In fact we run around 60-70% (replication).

I don't think right now it has to be spot on, since our ISP does allow bursting and some overages.  I just wanted some type of QoS in place to make sure we did not exceed 50 often so that we are not charged extra.

Thanks to all for contributing!

[Joseph W. Doherty]

Disclaimer

The     Author of this posting offers the information contained within this     posting without consideration and with the reader's understanding  that    there's no implied or expressed suitability or fitness for any   purpose.   Information provided is for informational purposes only and   should not   be construed as rendering professional advice of any kind.   Usage of  this  posting's information is solely at reader's own risk.

Liability Disclaimer

In     no event shall Author be liable for any damages whatsoever   (including,   without limitation, damages for loss of use, data or   profit) arising  out  of the use or inability to use the posting's   information even if  Author  has been advised of the possibility of  such  damage.

Posting

Then you'll likely be fine, as that sounds "run-of-the-mill".

If you were also dealing with VoIP traffic, then often you do need to be more "spot on".

Oh, and although you note you only "In fact we run around 60-70% (replication).", such can be deceptive since generally average usage is often measured across minutes.  Again, for your purpose to avoid extra bandwidth charges, this shouldn't be an issue.

PS:

If your provider allows bursting, and if they don't charge for it when you use it, you might also use shape peak vs. shape average.