cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
7533
Views
0
Helpful
6
Replies

Understanding QoS with Policy-Maps

mnlatif
Level 3
Level 3

Hi,

When using Policy Maps for implementing QoS, Is the below terminology correct

+++++++++++++++

policy-map mypolicy1

class myclass

+++++++++++++++

Under the Class, when using

Priority --> Mean strict queing and this much BW will Always be set aside and will not be available to non-matching traffic.Matching traffic, however can get more BW if available.

Bandwidth --> Router will try to make this much BW available to the matching application but its not Guaranteed, if the interface is fully congested. Matching traffic can get more BW than the specified value, if available.

Police--> No Min. BW is guaranteed. But router will limit the application to the specified value and drop any excess traffic, even when BW is available on the interface.

Thanks,

Naman

1 Accepted Solution

Accepted Solutions

Hi Naman,

Here are the differences:-

- the priority class is scheduled before any other queue (up to the configured bandwidth)

- under congestion conditions, the priority class cannot exceed the configured bandwidth, whereas the classes configured using bandwidth can.

So your understanding of that is correct.

For example, consider the following:

policy-map MyPolicy

class one

priority 128

class two

bandwidth 64

class three

bandwidth 128

If the above policy is applied to a 512k interface and the offered traffic for each class is as follows:

class one - 256k

class two - 256k

class three - 25k

the interface will be in a state of congestion.

The classes will be assigned approximate bandwidth as follows:

class one - 128k (policed sinced interface is congested)

class two - 64/(64+128)*(512-128) = 128k

class three - 128/(64+128)*(512-128) = 256k

The available bandwidth is shared in proportion between classes two and three.

Hope that helps - pls do rate the posts if it does.

Paresh

View solution in original post

6 Replies 6

ikizoo4
Level 1
Level 1

Hi Naman

Most of your statements are correct except about priority.

1. not always be aside : is used by non-matching traffic, because PQ and WFQ are performed before the Layer2 link based queueing.

2. PQ has internal policer which means PQ kbps parameteris always maximum bandwidth available to PQ.

Thanks

Hi,

There is no real concept of bandwidth reservation when using QoS on a cisco router. When you configure 'priority' and 'bandwidth' statements, the router deducts the specified bandwidth from the available bandwidth for the interface. This just means that you cannot assign bandwidths to classes in a manner that exceeds the interface bandwidth. This, however, is a static allocation. There is no hard bandwidth allocation on the interface to any traffic class. The actual bandwidth split happens when there is actual traffic on the interface.

In the context of the above, the following applies to the various types of QoS actions:

Priority - this is indeed a strict priority queue. You need to consider two cases when thinking about this. In a congested state, the priority traffic will be policed to the confifugred bandwidth and will not be allowed to exceed it. In a non-congested stated, the priority queue can use up any extra bandwidth available. Note that the bandwidth allocated to the priority class is not reserved. If the priority queue is not using up its full bandwidth, other classes are free to use that bandwidth.

Bandwidth - this is the minimum guarantee of bandwidth to the class. This class was always get at least this much bandwidth, even under conditions of congestion. If spare bandwidth is available on the interface, the class can get more than the specified bandwidth.

Police - your understanding of this is correct.

Hope that helps - pls do rate the post if it does.

Paresh

Thanks Paresh.

A little more clarification on Priority and Bandwidth. As per your definitation, i cannot see any apparent difference between two

1. Both can go above configured bandwidth, if available.

2. Both get atleast a MIN bandwidth as configured under congestion.

Then what is the basic difference ? Is it that with "bandwidth" the traffic is not really policed above the configured value during congestion ?

Thanks,

Naman

Hi Naman,

Here are the differences:-

- the priority class is scheduled before any other queue (up to the configured bandwidth)

- under congestion conditions, the priority class cannot exceed the configured bandwidth, whereas the classes configured using bandwidth can.

So your understanding of that is correct.

For example, consider the following:

policy-map MyPolicy

class one

priority 128

class two

bandwidth 64

class three

bandwidth 128

If the above policy is applied to a 512k interface and the offered traffic for each class is as follows:

class one - 256k

class two - 256k

class three - 25k

the interface will be in a state of congestion.

The classes will be assigned approximate bandwidth as follows:

class one - 128k (policed sinced interface is congested)

class two - 64/(64+128)*(512-128) = 128k

class three - 128/(64+128)*(512-128) = 256k

The available bandwidth is shared in proportion between classes two and three.

Hope that helps - pls do rate the posts if it does.

Paresh

Hi

I think Paresh is making a little mistake, please let me correct you with courtesy.

- the priority class is scheduled before any other queue (up to the configured bandwidth)

: which means priority class is not available more than configured bandwidth, that is right but not by scheduler, by internal policer on PQ, and this is incosistent with your previous statement that "priority class can exceed the configured bandwidth when available".

Like I said before, the configured bandwidth is maximum bandwidth available to PQ, PQ and other Queue are working seperately until reach to queue-scheduler on interface, so without policer at PQ, other queues always being starved.

I thinks you are confusing with IP RPT PRIORITY, it work like exactly what you saying,basically it is activated only when congestion happens.

my .02

Thanks

Thanks to All for help on this.

This is from Cisco website

+++++++++++++++++

When the device is not congested, the priority class traffic is allowed to exceed its allocated bandwidth. When the device is congested, the priority class traffic above the allocated bandwidth is discarded.

+++++++++++++++++

at

http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122cgcr/fqos_r/qrfcmd6.htm#wp1036073

So i think that Paresh is right about PQ exceeding the configured Bandwidth, when available.

Thanks \\ Naman

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: