Shaping will cause the traffic to adhere to the specific rate at which we have configured it even if there is available bandwidth to go faster. So in your example, the traffic would be limited to approximately 100 Mbps. Even if the queue limit is increased, the excess traffic still gets buffered even if there is no congestion on the same interface.
It is possible to configure shaping such that it can burst up to the speed of the physical interface. This is described in the following lesson:
With policing however, you are able to specify what happens if traffic exceeds the preconfigured settings where you can specify if it will be dropped or allowed. More about policing can be found here.
Okayā¦So lets say I am bursting it to full capacity of the interface for default class and there are already priority queues available to be served on the same physical interface so does those priority queues get affected if the bc and be for default class is configured equal to the cir??
It depends on how you configure this. For example:
policy-map PRIORITY
class class-default
priority 64
policy-map SHAPING
class class-default
shape average 48000
service-policy PRIORITY
Donāt mind the values, I just picked something. The shaper is the āoutsideā policy and the priority queue is the āinsideā policy. Your priority queue is affected by the shaper. Depending on the values you set, the shaper might shape the traffic of your priority queue or not.
Hello Guys,
I was trying out the Shaping labs and I noticed that as you mentioned Rene, my output for #show policy-map interface s0/1 was different to yours as my IOS is in fact an older version.
However, I am concerned even though after using iperf, I got results that were sometimes close to my CIR, the output says, āShaping Active No.ā Does this mean that even though I applied the service policy to the interface it is not shaping?
Thanks in advance, Regards
Take a look at this Cisco command reference. It states here that the indication under Shaping Active is the following:
Indicates whether the traffic shaper is active. For example, if a traffic shaper is active, and the traffic being sent exceeds the traffic shaping rate, a āyesā appears in this field.
So it looks like traffic must exceed the traffic shaping rate in order for a āyesā to appear here.
Also, you are correct that āSustain bits/intā = Bc and āExcess bits/intā= Be.
queue1-weight = Ratio of the low-priority queue weight; valid values are from 1 and 100 percent.
queue2-weight = (Optional) Ratio of the medium-priority queue weight; valid values are from 1 and 100 percent.
queue3-weight = Ratio of the high-priority queue weight; see the āUsage Guidelinesā section for valid values.
You can find out more information about this command at the following Cisco Command Reference.
Now if your question was what is the meaning of the queue limit value found in the output of the show policy-map interface command, then the answer to that is that this value shows the maximum number of packets that a queue can hold for the specific class policy configured in the policy map.
I tested ipref on my home/lab and Iāve discovered that the interface on my laptop (newer pc) is going to 100Mb/s.
On my cisco 3560CG I can see:
SW#show interfaces status
Port Name Status Vlan Duplex Speed Type
Gi0/1 connected trunk a-full a-1000 10/100/1000BaseTX
Gi0/2 connected trunk a-full a-1000 10/100/1000BaseTX
Gi0/3 mycomputer connected 10 a-full a-100 10/100/1000BaseTX
Gi0/4 connected 30 a-full a-1000 10/100/1000BaseTX
Gi0/5 notconnect 20 auto auto 10/100/1000BaseTX
Gi0/6 raspberry connected 30 a-full a-1000 10/100/1000BaseTX
Gi0/7 notconnect 30 auto auto 10/100/1000BaseTX
Gi0/8 connected 10 a-full a-1000 10/100/1000BaseTX
where my computer is connected to g0/3 and also I have a raspberry on g0/6.
Iāve tried to change interface / cable and also operating system ( I have dual boot )
Can you help me to understand this behavor?
why my nic is going a-100 ( iām not able to forze the speed ), mabye is
why my raspberry is going on 1Gb? ( I have Pi3 I know that it have only 100Mb )
The first thing that comes to mind, and the most probable answer from what I see, is that you have the raspberry connected to Gi0/3 and the laptop connected to G0/6. Although it is possible for your laptop to be functioning at 100Mbps, a slower speed than the NIC is rated, it is not possible to have the raspberry functioning at 1000Mbps, a higher speed than its NIC.
Now if this is not the case, then the next thing you should check out is what speed your laptop says it is connecting at. If youāre using Windows, you can check this from the Ethernet Adaptor settings:
If that is indeed at 100 Mbps, go to Properties ā Configureā¦ and go to the Advanced tab as shown below. Search for Speed & Duplex and check to see at what speed you are set at:
SW#show interfaces status
Port Name Status Vlan Duplex Speed Type
Gi0/1 TRUNK_OPNSENSE connected trunk a-full a-1000 10/100/1000BaseTX
Gi0/2 TRUNK_OPNSENSE connected trunk a-full a-1000 10/100/1000BaseTX
Gi0/3 - notconnect 10 auto auto 10/100/1000BaseTX
Gi0/4 - connected 30 a-full a-1000 10/100/1000BaseTX
Gi0/5 -mycomputer connected 20 a-full a-1000 10/100/1000BaseTX
Gi0/6 raspberry connected 30 a-full a-1000 10/100/1000BaseTX
Gi0/7 - notconnect 30 auto auto 10/100/1000BaseTX
Gi0/8 connected 10 a-full a-1000 10/100/1000BaseTX
Gi0/9 notconnect 1 auto auto Not Present
Gi0/10 notconnect 1 auto auto Not Present
Po1 connected trunk a-full a-1000
I only changed with another cable, now ipref is āworkingā ( I can go over 100Mb with my pc ) but even my raspberry is labled to 1gb but is going only 100Mbā¦
Hmm, Iām not sure what to say here, that is such a strange behaviour. Getting speeds of up to 100 Mbps for your raspberry looks like it is conforming to the correct speed, but it is still auto-negotiating to 1000Mbps, and that is strange. Are you connecting your raspberry directly to the switch or is there some other device in between? You mention an open sense firewall. Can you give us a little more information about your topology and how the FW fits in? It may also be helpful to hardwire the speed configuration on that particular interface to 100Mbps and experiment with 1000Mbps as well, just to see what happensā¦ Looking forward to hearing your results.
I think that the last time the bit rate was under 100Mbps because the packet should be routed by the firewall for inter-vlan routing configuration. (If so, maybe my FW has a 100Mbps portsā¦I will check thisā¦ )
Maybe USB-ethernet (like the raspberry interface ) adapters have different behavior during congestion than a traditional nic??
I think I cracked itā¦ It seems that this version of Raspberry Pi provides what they call āfaster Ethernetā as opposed to FastEthernet or GigabitEthernet. Apparently, based on this online review,
The Raspberry Pi 3 Model B+'s Gigabit Ethernet over USB 2.0 offers a maximum throughput of 300Mbps.
Which seems to jive with your results. But because the switch cannot connect at speeds of 300Mbps, it must either choose 100 or 1000Mbps. The electronics of the Ethernet port make the switch āthinkā that it is negotiating with a GigabitEthernet port, but the actual capabilities are limited to 300 Mbps. Keeping in mind that the theoretical maximum speed of a USB 2.0 port is at 480 Mbps, this speed seems reasonable.
Wanted to confirm I calculated a shaping policy I configured. The contracted rate is 200MB. The config is looking for bits per second. I scrubbed the names of the policy but you can the bits per second I entered. Is this the correct conversion?
policy-map xxx
class class-default
shape average 200000000
service-policy aaa
In your class map, you are indeed applying 200Mbps to the class-default class. This means that, without any additional classes defined, this will be the limitation on the interface the policy map is applied to.
Now the service-policy command seems to be out of place. This should be under the interface where you want to apply the policy map.
Hi Rene , I Have one question please :
if I implement the Sharping Policy for the default class Map as I need to match all traffic go through the Router. , is there any need for classifying the packets using access-list any more .
If you are planning to apply the shaping policy to all traffic, then no, you donāt need to classify packets using an ACL. This is because the shaping policy is applied to ALL TRAFFIC.
However, if you want to, at some point in the future, apply different policies for specific types of traffic, then you will need to create and use access lists to classify those packets accordingly.
So for futureproofing your configuration, it may be worth creating an access list that can later be modified or added to in order to achieve the kind of shaping you may want to achieveā¦
When weāre configuring shaping with CBWFQ, do we need to specify a shaping rate for each class or how exactly would this work? As it seems a little complicated to me to even think about it
)
