Does TR line card (ASR9K) support H-QoS?

pipatpong.samranpit · ‎10-11-2013

Hi,

I have some question regarding on the 2nd generation line card for ASR9k.

Line Card Types

--------------------

The Cisco ASR 9000 Series line cards are available in Service Edge Optimized and Packet Transport Optimized variants.

• Service Edge Optimized (SE) line cards are designed for customer deployments requiring enhanced quality of service (QoS).

• Packet Transport Optimized (TR) line cards are designed for network deployments where basic QoS is required.

Different line card types may be mixed within the same system.

I already review ASR9k presentation from Xander-Thuijs and found that

SE line card support 256K queues/NP

TR line card support 8 queues/Port

I'm not sure whether TR line card support H-QoS.

Could you please give more clarification about this?

Br,

Pipatpong

xthuijs · ‎10-13-2013

Yup, the TR card supports HQOS also, but you only have a limit number of queues per port available on those cards.

Egress policy consume a queue when you use shape, bandwidth or priority.

In that same presentation you are looking at there is an example command that shows you how to find out if your

config burns a queue or not. Every unique QID is a minus one from the 8 you have per port/phy interface.

So if your QOS pmap, hierarchical or not, burns 4 queues, then you can apply it to 2 subinterfaces before it starts

throwing errors.

regards

xander

pipatpong.samranpit · ‎10-13-2013

Hi Xander,

Thank you so much for your explaination.

Which IOS XR command use to verify the configuration burns a queue or not? Please explain about this command and the result?

I'm not sure wheather you mentions about the "show policy-map interface" command. Am I correcty understood?

Br,

Pipatpong

xthuijs · ‎10-13-2013

HEre is an example:

RP/0/RSP0/CPU0:A9K-BNG#show qos int g 0/0/0/0 out

Sun Oct 13 13:31:19.867 EDT

Interface: GigabitEthernet0_0_0_0 output

Bandwidth configured: 200000 kbps Bandwidth programed: 200000 kbps

ANCP user configured: 0 kbps ANCP programed in HW: 0 kbps

Port Shaper programed in HW: 200000 kbps

Policy: xtp Total number of classes: 4

----------------------------------------------------------------------

Level: 0 Policy: xtp Class: class-default

QueueID: N/A

Shape CIR : NONE

Shape PIR Profile : 0/4(S) Scale: 195 PIR: 199680 kbps PBS: 2496000 bytes

WFQ Profile: 0/9 Committed Weight: 10 Excess Weight: 10

Bandwidth: 0 kbps, BW sum for Level 0: 0 kbps, Excess Ratio: 1

----------------------------------------------------------------------

Level: 1 Policy: xt Class: m1

Parent Policy: xtp Class: class-default

QueueID: 136 (Priority 1)

Queue Limit: 4 kbytes Abs-Index: 4 Template: 0 Curve: 6

Shape CIR Profile: INVALID

Policer Profile: 53 (Single)

Conform: 296 kbps (300 kbps) Burst: 3750 bytes (0 Default)

Child Policer Conform: TX

Child Policer Exceed: DROP

Child Policer Violate: DROP

----------------------------------------------------------------------

Level: 1 Policy: xt Class: m2

Parent Policy: xtp Class: class-default

QueueID: 137 (Priority 2)

Queue Limit: 8 kbytes Abs-Index: 6 Template: 0 Curve: 6

Shape CIR Profile: INVALID

Policer Profile: 54 (Single)

Conform: 600 kbps (600 kbps) Burst: 7500 bytes (0 Default)

Child Policer Conform: TX

Child Policer Exceed: DROP

Child Policer Violate: DROP

----------------------------------------------------------------------

Level: 1 Policy: xt Class: class-default

Parent Policy: xtp Class: class-default

QueueID: 138 (Priority Normal)

Queue Limit: 2496 kbytes Abs-Index: 89 Template: 0 Curve: 0

Shape CIR Profile: INVALID

WFQ Profile: 0/9 Committed Weight: 10 Excess Weight: 10

Bandwidth: 0 kbps, BW sum for Level 1: 0 kbps, Excess Ratio: 1

----------------------------------------------------------------------

This class default will get reused by other interaces very likely unless you have a bw remaining (percent) in there.

So every time you see a unique QueueID you burn a queue.

the reason why the parent shaper at level 0 says N/A is because this is a hierarchical pmap and is shaping the overal chuld classes inclduing class-default.

So in this example applying this pmap multiple times will burn queues for class-default.

In other words, this Pmap example uses 3 queues. One for the shaper (child class-default) and 2 PQ's of prio's 1 and 2).

policy-map xtp

class class-default

service-policy xt

shape average 200 mbps

!

end-policy-map

!

RP/0/RSP0/CPU0:A9K-BNG#sh run policy-map xt

Sun Oct 13 13:34:45.049 EDT

policy-map xt

class m1

priority level 1

police rate 300 kbps

!

class m2

priority level 2

police rate 600 kbps

!

class class-default

!

end-policy-map

pipatpong.samranpit · ‎10-14-2013

Hi Xander,

Every pmap will always has class "class-default even if we don't define it in pmap. It is default class in pmap.

So, that means we have to dedicate one queue for it. Am I correcty understood?

Thank you so much in advance.

Br,

Pipatpong

xthuijs · ‎10-14-2013

Lets say if you have 2 sub interfaces off the same phy, then if you have 2 parents child policies with parent shapers,

then both class defaults will be unique.

xander

pipatpong.samranpit · ‎10-14-2013

Hi Xander,

Thank you so much for your explaination.

You give quick and smart answers for every questions.

Br,

Pipatpong

sbhadrav@cisco.com · ‎05-18-2018

Both MOD80s (SE and TR) have two of the same NP, so in terms of raw bandwidth no, no difference.

Saying this another way, if the same configuration is applied to both cards you should see the same performance. Which card you choose to use really comes down to the needs of your network..

smilstea · ‎05-18-2018

Not entirely true. The cards differ in terms of memory, CPU, and TCAM so the same config might not work on both cards due to hardware limitations. If you plan to use a lot of ACLs or QoS policies then the SE variant is likely a better choice. Also certain features like BNG require the SE variant card.

Sam