question about QOS settings

Hello all. I have a 3AP 900mhz tower with 55, 45 and 30 users on it. Control slots currently set to 1, downlink % set to 75. I currently have QOS settings:

sustained downlink and uplink the same
burst down and up the same.

I was browsing through older posts about control slots and found a similar post at: viewtopic.php?f=1&t=6602&hilit=control+slots

And found that I may be wasting air time b/c of my QOS settings. Is this so? Should I set my up settings to 25% of my downlink? I want to get every drop of performance out of these APs before I have to upgrade/add.

I have tried changing my control slots to as much as 4 but didn’t see any improvements.

Thank you.

keep them no more than 2, it is for canopy traffic assosations ect. no need to have it higher, we leave ours at the default 0 and no issues, our managment software shows we are getting full performance from the radios and latency says good.

Seriously? the docs have a chart for the control slot settings. I may try it to see.

Also, have you heard of setting your max range to twice the distance of your furthest customer if you want to use 2x? I read that on another forum. I currently have mine set to 1/2 mile over my furthest sub.

I have not heard of that either…but, I am having some throughput issues with my AP, and they are not even close to the maximum throughput they are supposed to be capable of. I need some advice on this.

Yes, with QoS settings like that you are probably wasting air time.

You will want to set your DL and UL rates to match your Downlink Ratio. In our case we have 75% Downlink Ratio and we sell 1500/512 kbps service. As I understand Canopy’s bandwidth allocation, 25% of the airtime is for upload. Of course I could be wrong and the reality might be that the upload is 25% of the download…

In any case, you will definitely want to match your QoS on both. I run with the rate I quoted above for 2X clients and a reduced rate of 1024/256 kbps for 1X clients. This keeps them from chomping all the available airtime, as if the 1X clients had the same QoS they could slaughter the APs. I also have a reasonable burst in - about 1 MB down and 0.5 MB up. This is working great for me on an AP with ~45 clients. If you sell slower speeds, than even better! If you don’t sell speeds, well then, play around with burst size and sustained rates until you find something that fits.


I have not heard of max range having any effect on being able to maintain a 2X link… I’m not an RF engineer but I don’t see how it could help since 2X is dependent on basic link quality (as reflected by your jitter) since it just turns up a timer in the AP so it listens for longer. I could see that helping in scenarios with multipath, but not 2X

Downlink% deals with the percentage of traffic you expect to be download (from the SM’s point of view). Yes, if you don’t set this right you might have trouble delivering speeds promised, but it also doesn’t make sense to set it to 50% unless 50% of your traffic is indeed upload traffic, otherwise you’ll be wasting air time.

Max distance has no effect on 1x/2x performance. It is as salad says, just a parameter that effects how long an AP will wait for an SM. Everything in Canopy runs on timing.

Use the frame calculator to help you determine the best settings for your deployment - it will allow you to figure out if you are wasting air time or not (they give you a downlink and uplink half slots stat - if this is not cleanly divisible by 2, you’ve lost a frame somewhere and are not getting the max potential from your AP). The other thing you will want to do is make sure that in every circumstance that you can possibly control, your SM is in 2x. The AP operates twice as efficient in 2x, which gives you extra aggregate speed for all SMs. If you, however have a mix of say 50/50 of 1x and 2x, you’ll never reach the full 2x rate of the AP because those 1x SMs use the same amount of air time, but can only transmit half as much data.

My practice is to allocate a little bit of burst to all my SMs (the burst being increased with each subsequent service package). The idea is to get those users on and off the network with their traffic as quickly as possible. 1 user downloading a file at 1mbps or uploading a file at 512kps for sustained periods of time could completely saturate a 900MHz AP relatively easily, affecting performance for everyone on that AP. You’ll see some Canopy operators that market based on their burst speeds and have lower sustained speeds (for the really long file transfers), but it’s all in how you want to market your service and the habits of your subscriber base.

Hope that all makes sense.

I have noticed with my latest sms running 9.4.x that they have an adaptive rate option (1x/2x). So, if the SM is great at 1x, is it safe to switch it to this instead of locking it on 1x? I realize that if it is not a great connection, it will error correct as much as it can, but since it says adaptive, it will switch between modes automatically, right?

Is there any harm in setting all sms to this setting?

The case can be argued that even if an SM is only in 2x mode 10% of the time, it’s still more efficient than being in 1x mode 100% of the time. 50% linktests in 2x is essentially 1x mode.

I run most of my links in 2x. I do come across SMs here and there that just do not meet spec and will keep them in 1x, but I think it would be alright to run everything in 2x dynamic rate switching.

+1

wifiguy wrote:
I run most of my links in 2x. I do come across SMs here and there that just do not meet spec and will keep them in 1x, but I think it would be alright to run everything in 2x dynamic rate switching.
Not sure about this. If one of SM´s connected to the AP is set to 1x only, the ap throughput will be half?
kejsix wrote:
Not sure about this. If one of SM´s connected to the AP is set to 1x only, the ap throughput will be half?


Not quite - the issue is that moving data at 1X takes roughly twice the amount of airtime as moving it at 2X. If you have a lot of 1X clients moving data this chews up airtime on the AP in a hurry.

Say you have a 900 MHz AP with four radios on it. Two in 2X mode and two in 1X mode. Let's assume all are perfect links and we'll aggregate the bandwidth.

Single 2X client: 4 Mbps total; 4 Mbps per client
Single 1X client: 2 Mbps total; 2 Mbps per client
Two 2X clients: 4 Mbps total; 2 Mbps per client
Two 1X clients: 2 Mbps total; 1 Mbps per client
Split 1X/2X: 3 Mbps total; 1 Mbps for the 1X client, 2 Mbps for the 2X client
Split 1X/2X/2X: 3.33 Mbps total: 0.67 for the 1X client, 1.33 Mbps for the 2X clients
Split 1X/1X/2X: 2.67 Mbps total: 0.67 for the 2X clients, 1.33 Mbps for the 2X client
Split 1X/1X/2X/2X: 3 Mbps total; 0.5 Mbps for the 1X clients, 1 Mbps for the 2X clients

These numbers are the same (in terms of aggregate bandwidth from the AP) as long as you keep the same ratio.

Simplified further, an AP with 100% 2X links can provide twice the quality of service/twice the bandwidth/carry twice as many customers/make you twice as much money as an AP with 100% 1X links. However the reality falls somewhere in between.

Hope this helps