cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
24441
Views
24
Helpful
9
Replies

Understanding mls qos srr-queue input threshold command

ssieger
Level 1
Level 1

Hi,

can somebody explain me the meaning of the two parameters threshold-percentage1 and threshold-percentage2 of the command mls qos srr-queue input threshold? What is the effect, an when is it recommended to modify this?

Tia,

Stephan

1 Accepted Solution

Accepted Solutions

If thresold 1 in crossed over 50 percent, it starts dropping packets known as SPD to avoid congestion ( e.g. it drops 1 out of 10). If the input still grows say to 80 it drops packet ( e.g drops 6 out of ten) more thans what it was dropping when it crossed 50. When it reaches 100 it drops every packet.

HTH

Hitesh Vinzoda

Please rate useful posts.

View solution in original post

9 Replies 9

podhillo
Level 1
Level 1

This example shows how to configure the tail-drop thresholds for the two queues. The queue 1 thresholds are 50 percent and 100 percent, and the queue 2 thresholds are 70 percent and 100 percent:

Switch(config)# mls qos srr-queue input threshold 1 50 100

Switch(config)# mls qos srr-queue input threshold 2 70 100

QoS uses the CoS-to-threshold map or the DSCP-to-threshold map to decide which class of service (CoS) or Differentiated Services Code Points (DSCPs) values are mapped to threshold 1 and to threshold 2. If threshold 1 is exceeded, packets with CoS or DSCPs assigned to this threshold are dropped until the threshold is no longer exceeded. However, packets assigned to threshold 2 continue to be queued and sent as long as the second threshold is not exceeded.

Each queue has two configurable (explicit) drop threshold and one preset (implicit) drop threshold (full).

You configure the CoS-to-threshold map by using the mls qos srr-queue input cos-map global configuration command. You configure the DSCP-to-threshold map by using the mls qos srr-queue input dscp-map global configuration command.

Hi podhillo,

that's what i found on the cisco website, but what happens when 50 percent of the threshold 1 ist reached. If it starts dropping what is the meaning of the 100 percent parameter? That is not clear for me.

Greetings,

Stephan

If thresold 1 in crossed over 50 percent, it starts dropping packets known as SPD to avoid congestion ( e.g. it drops 1 out of 10). If the input still grows say to 80 it drops packet ( e.g drops 6 out of ten) more thans what it was dropping when it crossed 50. When it reaches 100 it drops every packet.

HTH

Hitesh Vinzoda

Please rate useful posts.

Hi,

that's it. As i read your post i call it to my mind, many thanks.

Greetings,

Stephan

To my opinion it works a bit different. You have 2 input queues, each with 3 thresholds. The third is not configurable, it is the 100% of the buffers available to the queue threshold. Thresholds 1 en 2 can be assigned a value. These values are also a percentage of the total buffers available to the queue, however mostly somewhat lower.

Incoming traffic is matched against these values based on their COS or DSCP value, configured by these commands:

mls qos srr-queue input cos-map queue 1 threshold 2 1

mls qos srr-queue input dscp-map queue 2 threshold 3 40 41 42 43 44 45 46 47

What happens is that a frame comes in an interface, the switch looks at its COS value and matches it against the configured thresholds. Lets say it matches against queue 1 threshold 1 which is configured to 25%, if the the current ingress buffer use is lower than 25% the packet is queued, if it is higher the packet is dropped.

You can for example configure threshold 2 for queue 1 to 66%. This means traffic which is mapped to this queue and threshold is less likely to get dropped. Traffic mapped to threshold 3 will only be dropped when the ingress buffers are fully utilized.

Via this way you can give different COS values different levels of "importance". So when you now assign default COS to threshold 1 and VoIP (CS5) COS to threshold 2, you have given your voice traffic some priority over other traffic.

This is the correct answer. Please update.

Thanks man, perfect answer ! 10x

I now figure out how this logic works

So something like first threshold is minimum range when to start drop and second threshold is where to drop all packets like tail drop :) and in range from minimum to maximum grow dropping logic ) 

Thank you again )  

I believe for SRR WTD drops, the drops limits are absolute for what they match.

For example, using AFxy, you might drop all AFx3 at 50%, all AFx2 at 75%, and remaining AFx1 at 100%.

Joseph I want to catch exact and true logic, so here in attached pic, I have in queue 1 threshold  1 - 80 and threshold 2 - 90  as we see on pic, and queue 2 thres 1 100 and thres 2 100 

let's say we have next this command:

mls qos srr-queue input cos-map queue 1 threshold 2 6 7

 So, this means that when input packets came with CoS label 6 and 7 will appear with threshold 2 drop range (at 90% load will start dropping and at 100 will full drop) if not than go in queue 1

if I will have mls qos srr-queue input cos-map queue 2 threshold 1 3 4 5 

cos label 3,4 and 5 will appear with threshold 1 drop range will start at 80 %  and full drop will in 100 % go in 2 queue. 

so with this logic upper threshold is non-configurable too :) 

Have I healthy logic or not yet ? 

threshold 3 is non-configurable and have range 100 % like tail drop not so like but similar ) 

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:

Review Cisco Networking products for a $25 gift card