Cisco QoS Minimum and Maximum Bandwidth

[sblakely]

I'm trying to set up a QoS policy-map on a Cisco 9300L that replaced an existing Juniper QoS policy. Ideally I want to match it exactly but I'm running into a basic difference between the two. Many of the Juniper policies have a transmit-rate and shape-rate applied to them. The transmit-rate defines a minimum allocated bandwidth during congestion and the shape-rate defines a maximum average bandwidth through the link. My understanding is there are similar options with Cisco, the bandwidth and shape commands, but attempting to add both of them to a single class within a policy-map returns an "unsupported" error. Is there any method to applying a similar setup as the Juniper CoS on Cisco with a minimum allocated bandwidth during congestion and a maximum bandwidth both on the same class?

[Joseph W. Doherty]

Unsure a 9300L supports the combination; possibly why you're obtaining an unsupported error message.

What IOS is in running on the 9300L?

[sblakely]

CAT9k IOS XE 17.13.01

[Joseph W. Doherty]

From a quick read of the 9300 17.13.x configuration guide, it appears shape and bandwidth are supported within the same class.  There is a restriction that all classes must use the same bandwidth kind of parameters, not noted is using different parameter kinds for the shape and bandwidth statement.

Didn't find any detailed information on bandwidth or shaper statements in 17.13.x command reference.

Did find a few QoS restrictions in the 17.13.x release notes - none relevant (I think).

Possibly some restriction with the "L" model, or feature license, or whatever you're actually trying to do.  For the latter, can you post relevant class-maps, policy-maps, interface config before service policy addition, and the exact output when you try to apply policy-map, both at CLI and any corresponding syslog messages?

[MHM Cisco World]

https://www.cisco.com/c/en/us/support/docs/quality-of-service-qos/qos-packet-marking/10100-priorityvsbw.html

check this link 

MHM

[Joseph W. Doherty]

---

@MHM Cisco World wrote:

https://www.cisco.com/c/en/us/support/docs/quality-of-service-qos/qos-packet-marking/10100-priorityvsbw.html

check this link 

---

Is your suggestion to police rather than shape?

If so, if OP believe policing would also be a satisfactory approach, it's worth trying, as policing doesn't need to create queues as shaping does.

As a side note, Cisco's general QoS documentation, like the above referenced, continues to improve!  This particular documentation also describes the tx-ring-limit and the possible need to adjust it when supporting VoIP (true!), the need to account for L2 overhead, and also describes how LLQ policing usually only comes into effect when there's actual congestion.

Further, QoS configuration feature options also continue to improve!  Shortly ago, just reading how on the particular IOS I was reading, it can set policy-map bandwidth allocations as ratios to each other (an approach I've been doing, and suggesting for years, using percentages).

[JeffLawson]

---

@sblakely wrote:

I'm trying to set up a QoS policy-map on a Cisco 9300L that replaced an existing Juniper QoS policy. Ideally I want to match it exactly but I'm running into a basic difference between the two. Many of the Juniper policies have a transmit-rate and shape-rate applied to them. The transmit-rate defines a minimum allocated bandwidth during congestion and the shape-rate defines a maximum average bandwidth through the link. My understanding is there are similar options with Cisco, the bandwidth and shape commands, but attempting to add both of them to a single class within a policy-map returns an "unsupported" error. Is there any method to applying a similar setup as the Juniper CoS on Cisco with a minimum allocated bandwidth during congestion and a maximum bandwidth both on the same class?

---

In Cisco, to achieve both minimum and maximum bandwidth on a single class within a policy-map, you can use the bandwidth command for ensuring a minimum allocated bandwidth during congestion and the police command to set a maximum bandwidth limit. This combination allows you to manage both bandwidth guarantees and caps effectively within the same policy.