Solved: What does "shape average percent" mean in the QOS policy

Kevin Melton · ‎01-04-2010

I am working at a client site that is an MPLS customer. The customer has an MPLS circuit that runs between their Main HQ and their Disaster Recovery site. I have been asked to analyze and report as well on the way the Qos Policy is written, and to provide any recommendations on how they can improve performance.

There is a statement within the Qos Policy as it exists at each end on the 3825 routers. The statement is called "shape average percent". Here is the policy from one side:

policy-map QoS

class COS2_traffic

set dscp af31

shape average percent 12

bandwidth percent 13

class COS3_traffic

set dscp af21

shape average percent 27

bandwidth percent 28

random detect dscp-based

class class-default

fair-queue

random detect

What does this statement mean and how is it different than the the "bandwidth percent" statement?

Thanks

Kevin

Jon Marshall · ‎01-04-2010

k-melton wrote:
I am working at a client site that is an MPLS customer.  The customer has an MPLS circuit that runs between their Main HQ and their Disaster Recovery site.  I have been asked to analyze and report as well on the way the Qos Policy is written, and to provide any recommendations on how they can improve performance.
There is a statement within the Qos Policy as it exists at each end on the 3825 routers.  The statement is called "shape average percent".  Here is the policy from one side:
policy-map QoS
  class COS2_traffic
   set dscp af31
    shape average percent 12
    bandwidth percent 13
  class COS3_traffic
    set dscp af21
     shape average percent 27
     bandwidth percent 28
     random detect dscp-based
  class class-default
     fair-queue
      random detect
What does this statement mean and how is it different than the the "bandwidth percent" statement?
Thanks
Kevin

Kevin

The bandwidth percent is a minimum bandwidth guarantee so if there is spare bandwidth that can be used from other classes it will be used.

The shape average percent is a maximum bandwidth guarantee so any packets above that bandwidth will be queued and sent when there is spare bandwidth within that class.

So they are doing 2 fundamentally different things.

Jon

View solution in original post

Jon Marshall · ‎01-04-2010

k-melton wrote:
I am working at a client site that is an MPLS customer.  The customer has an MPLS circuit that runs between their Main HQ and their Disaster Recovery site.  I have been asked to analyze and report as well on the way the Qos Policy is written, and to provide any recommendations on how they can improve performance.
There is a statement within the Qos Policy as it exists at each end on the 3825 routers.  The statement is called "shape average percent".  Here is the policy from one side:
policy-map QoS
  class COS2_traffic
   set dscp af31
    shape average percent 12
    bandwidth percent 13
  class COS3_traffic
    set dscp af21
     shape average percent 27
     bandwidth percent 28
     random detect dscp-based
  class class-default
     fair-queue
      random detect
What does this statement mean and how is it different than the the "bandwidth percent" statement?
Thanks
Kevin

Kevin

The bandwidth percent is a minimum bandwidth guarantee so if there is spare bandwidth that can be used from other classes it will be used.

The shape average percent is a maximum bandwidth guarantee so any packets above that bandwidth will be queued and sent when there is spare bandwidth within that class.

So they are doing 2 fundamentally different things.

Jon

Santosh Shetty · ‎06-19-2012

I have a query, in above example how these command workd together

shape average percent 27 << is this 27% of total BW available?

bandwidht percent 28

As per the above explanation it's look like Min BW allocated is 28% and Maximum 27%. Should it not be Min BW value lower than Max BW value?