We are looking to implement Diffserv QoS within our 450i infrastructure
version 16.0.1 is the version that we are going to upgraded to be for implementing QoS
basicaly what we are looking into acheiving is marking the traffic on the routers to we the AP\SM are connected and police it on the PMP
Could anyone pleae share any availble guide\documentation on how the traffic is matched on the SMs\APS against the different priority queues and share their experience in implementing this, any issues that we need to be aware of in terms of QoS on 450i on version 16.0.1
the Release note documentation is not the best that you can eget for this specific subject
I'll try to be as brief as possible here but with as much information as possible. Please note, the following information applies to Canopy 15.2 and later software. Looking at the Diffserv table on a PMP radio, you can see the default settings look like this:
The left-most column is fixed and cannont be changed, but all other code points use standard defaults and can be modified to suit your needs. Also, note that CP46 defaults to 802.1p CoS level 5, which is used for VoIP and will always map to High priority level (when 2 or more priorites are enabled), unless the device has only Low priority enabled.
Next, in terms of mapping based on number of priority levels, the mapping is as follows:
# of Priority Levels / Data Channels ------>
802.1p CoS 1 2 3 4
COS0 LOW LOW LOW LOW
COS1 LOW LOW LOW LOW
COS2 LOW LOW MED MED
COS3 LOW LOW MED MED
COS4 LOW HIGH HIGH HIGH
COS5 LOW HIGH HIGH HIGH
COS6 LOW HIGH HIGH U. HIGH
COS7 LOW HIGH HIGH U. HIGH
As you can see, depending on how many Data Channels (i.e. levels of priority) you have enabled on a given device, the mapping of Codepoints-->802.1p CoS--->Canopy priority levels changes. We deviated slightly from standard when 3 Data Channel priorities are in use to ensure that COS5 is always mapped to High priority when it's available. This is to ensure on 450m devices that the VoIP traffic is not grouped for MU-MIMO by default, which can introduced some latency for the benefit of higher sector throughput.
Also note that if one SM is only given a single Low Priority Data channel, while another SM is given a Low and High Priority data channel, for example, the first SM's 802.1p CoS 4-7 will be considered Low priority, while the 2nd SM's CoS 4-7 will be considered High priority. This means the second SM's traffic for those CoS levels would be prioritized higher by the Canopy link, but would be treated the same at other parts of your network.
Many thanks for you reply it was indeed very informative !
So in fact we have 4 different classes of service and not 8 as there are only 4 data channels , is this a correct statement ?
Another very important factor which will determine the outcome of the QOS operations is the available bandwidth which in this case is not a constant (factor to link quality); when I look into the QoS implementation on the SM and the AP the bandwidth is set staticly which I think will lead into unimpredicted outcome when the availble bandwidth to an SM is degraded
Idealy, it would have been more efficient to have percentages based on the dynamicaly availble bandwidth against the different classes instead of a fixed amount of badnwdith ( I appreciate how complicated implementing such a feature and the amount of processing resources required for it ...just a thought :) )
At the moment we are managing the QoS operation on Cisco routers , the reason for which we are looking into moving them to Cambium is that there is no way that Cisco routers will know for sure the exact availble bandwidth;
but it looks like the same issue applies to the Cambium devices themselves, is that the case ? is yes, do you agree on the fact that the end result of applying the QoS on the Routers or the PMP will be the same ? I'v eread on the documentation that WFQ can be used without CIR I'm not sure to fully understand how this could be done , Any chance that you coudl explain ?
What's the best way to check if the QoS is performing as expected I can see that we can have some statistics under statics ->Data Channels; is there any other tools \ options to check QOS operations
Many thanks for you reply it was indeed very informative !
So in fact we have 4 different classes of service and not 8 as there are only 4 data channels , is this a correct statement ?
Yes, as far as the AP-->SM link is concerned, Canopy supports up to 4 levels of user data CoS. However, standard protocols like DiffServ are handled as expected outside of that airlink for compatibility with your other network equipment.
Another very important factor which will determine the outcome of the QOS operations is the available bandwidth which in this case is not a constant (factor to link quality); when I look into the QoS implementation on the SM and the AP the bandwidth is set staticly which I think will lead into unimpredicted outcome when the availble bandwidth to an SM is degraded
Idealy, it would have been more efficient to have percentages based on the dynamicaly availble bandwidth against the different classes instead of a fixed amount of badnwdith ( I appreciate how complicated implementing such a feature and the amount of processing resources required for it ...just a thought :) )
At the moment we are managing the QoS operation on Cisco routers , the reason for which we are looking into moving them to Cambium is that there is no way that Cisco routers will know for sure the exact availble bandwidth;
but it looks like the same issue applies to the Cambium devices themselves, is that the case ? is yes, do you agree on the fact that the end result of applying the QoS on the Routers or the PMP will be the same ? I'v eread on the documentation that WFQ can be used without CIR I'm not sure to fully understand how this could be done , Any chance that you coudl explain ?
You are correct that Weighted Fair Queuing (WFQ) is meant to help handle what you described with respect to link quality and available bandwidth. What WFQ provides is a way to allocate a percentage of available "slots" to each class of service. While this means the actual bandwidth to each priority level can fluctuate with link quality, it also means that "airtime" or number of slots used won't cause any given priority level to hog all the available airtime.
Click Documentation, then under the tab for "Guides" select the link for "PMP/PTP 450 Series User Guide (pmp-2384)". The sections in Chapter 7 about configuring QoS should answer a lot of your questions.
