QOS

Hi guys. This is definately a noob question. Im used to the pmp100 system and its features and im slowly migrating to the epmp1000.

 im not sure how to adjust the speed settings. Currently its set to default and everyone is getting full speed which is ok right now because the only customers on the new system are business that requested the speed. Pretty soon ill be putting residental customers on it but want to set limits. Any advice is greatly appreciated :)


@geraldthunder wrote:

Hi guys. This is definately a noob question. Im used to the pmp100 system and its features and im slowly migrating to the epmp1000.

 im not sure how to adjust the speed settings. Currently its set to default and everyone is getting full speed which is ok right now because the only customers on the new system are business that requested the speed. Pretty soon ill be putting residental customers on it but want to set limits. Any advice is greatly appreciated :)


Hi, 

Welcome to ePMP! 

You can limit the speed that each SM gets using the MIR (Maximum Information Rate) feature under Configuration->Quality Of Service on the GUI. On the AP,  you can create the MIR profiles with the Downlink and Uplink speeds for that profile/package. Then you go to the SM and configure the MIR profile you want that particular SM to use. This configuration can also be done via RADIUS. 

Thanks,

Sriram

2 Likes

Could you please explain how this can be achieved using radius ?


@gregoryeid1 wrote:
Could you please explain how this can be achieved using radius ?

Hi, Please see this Knowledge Base article on how to configure MIR using RADIUS: http://community.cambiumnetworks.com/t5/ePMP-Configuration-Management/Configuring-Maximum-Information-Rate-MIR-Profiles-on-a-RADIUS/m-p/45297#U45297

and let us know if you have further questions. 

Thanks,

Sriram

1 Like

Fantastic. Very helpful! Thanks

Is there a way to enforce fair usage policy by changing the CPE’s MIR when data usage reaches a daily or weekly threshold?


@gregoryeid1 wrote:
Fantastic. Very helpful! Thanks

Is there a way to enforce fair usage policy by changing the CPE's MIR when data usage reaches a daily or weekly threshold?

You're welcome. Unfortunately, we don't have a way to change the MIR rate on the radio based on usage. Perhaps you could use a backend device to do this?

Hopefully other community members will chime in on how they solve this challenge. 


@gregoryeid1 wrote:
Fantastic. Very helpful! Thanks

Is there a way to enforce fair usage policy by changing the CPE's MIR when data usage reaches a daily or weekly threshold?

As Sri mentioned, there are many ways on the backend to do this. We use a QoS/DPI appliance from SandVine.

2 Likes

Thanks Eric for the feedback. We have tried many QoS/DPI appliances over the years. Allot, Packeteer, Emerging Technologies, MicroTik, and more.

We did not apply any QoS on Cambium last mile and allowed all traffic to hit the QoS/DPI appliance, which was at our core. The appliance was used to set MIR by source IP and priority by application. 

We found that said appliances performed poorly in many ways. Some had a queue size limit. Others did not have DPI capabilities. Appliance resources suffered (CPU mainly).

We have struggled to get on top QoS management of our network & customer traffic. Maybe we are expecting too much from an appliance. Maybe we should adopt a mixture of QoS strategies (QoS on CPE + DPI on core appliance, etc.)

Does this sound familiar? are we the only ones challenged by QoS? any help will be greatly appreciated.

1 Like

For the ePMP gear, we created a list of all the MIR settings we wanted to be able to support, and programmed the same set into each AP.  (0=default=4/1, 1=7/1, 2=10/2,  etc with 9=unlim/unlim)  We program every SM to MIR=0 by default, when a customer pays for more bandwidth we set this appropriately.

We do NOT currently impose any usage limits on our customers, so this is pretty much it.

We have, however, considered automatically limiting customers who are overtly hogging sector resources.  The mechanism I was considering using was simply to change (from our monitoring server, via script using SNMP) an identified abuser's SM priority to 'low' instead of 'normal'.  (we reserve 'high' for hand-tweaked setup on a very limited number of VIP customers)  That way they would still be able to get their 4mbps/7mbps/10mbps/whatever downlink all day long, UNLESS it interfered with other customers set to 'normal' or 'high' priority.

I've always favored the KISS principle, and this seemed the simplest and least obtrusive way to control the handful of customers who tend to hammer the sector for hours on end.  Changing their actual MIR setting would mean some external (to the radio) mechanism is needed to track what MIR they /were/ on so it could be restored in the future.  Just kicking them down to 'low' priority is quick and easy, and if 'low' is otherwise unused it makes it very easy to identify customers who have been limited this way and unbottle them manually or automatically later.

j

3 Likes