PHY Errors, Noise Floor, Airtime interpretation in ZD

  • 4
  • Question
  • Updated 2 years ago
I would like to find out what values for PHY Errors, Noice Floor, Airtime total, busy, RX, TX are good and what values are bad?

If I'm correctly informed, this is what I know:
PHI Errors of < 20000 is acceptable,
Noise Floor above -85 is bad, below -110 is excellent.
Total Airtime 100% would be great since that means the capacity is fully used, total airtime of < 20 could mean there is a lot of interference, or hardly any users.

Is this correct?
What can I learn from the other Airtime values (busy, rx, tx)? Are the values different for 2.4 and 5 GHz? (i.e 2.4GHz NF should be below -90, 5GHz NF should be below -85?)

I'm basically looking for an answer on this question: what values of which fields of the monitoring page in the ZD indicate that an AP is doing great or needs attention?
Photo of Jelle Alten

Jelle Alten

  • 60 Posts
  • 12 Reply Likes

Posted 3 years ago

  • 4
Photo of Primož Marinšek

Primož Marinšek, AlphaDog

  • 413 Posts
  • 48 Reply Likes
You've asked a great question and I hope for a nice discussion on this.

It's a bit subjective and it's not quite clear cut what is acceptable and what not. Maybe just one of many clients will cause most of the errors and others will do less and work fine.

But in any case I would not consider PHY errors of s a good compromise between price/performance.
Photo of the Event Engineers

the Event Engineers

  • 7 Posts
  • 1 Reply Like
Hi Jelle ,

Here are our numbers as seen on a 7762AC deployed indoors, during a tech conference. At the moment we recorded these numbers, it was almost impossible to get a reliable connection on the 2.4 GHz frequency.

Radio: 802.11g/n
# of Aurhorized Client Devices: 69
Noise Floor: -88
PHY Errors: 831
% AirTime (total/busy/RX/TX): 87/47/33/8

%Retries/%Drops: 7.14 / 0.00
TX Power: Min
Background scanning: enabled
Photo of Primož Marinšek

Primož Marinšek, AlphaDog

  • 413 Posts
  • 48 Reply Likes
Well you've got low retry rate which is good, also your Air Time is at 87% but the Air Time declared as busy is at 47% and the noise floor is at -88dBm.

I guess there are quite a few other (or your) networks around your AP that keep triggering this APs and the STAs connected to it CCA.

This could also be from people bringing their MiFis. How many networks do are there?

You could try using a higher directional antenna or 7762-AC-S and stop transmitting 802.11b rates. Those travel far and cause a lot of issues. You do that over CLI.
Photo of Jelle Alten

Jelle Alten

  • 60 Posts
  • 12 Reply Likes
So if I zoom in on Airtime:
High "total" is good, if "busy" is not a big part of it?
Or even: if busy < rx + tx, the AP is doing ok?
Photo of Jelle Alten

Jelle Alten

  • 60 Posts
  • 12 Reply Likes
Uh uh, now I have an AP that has no clients, but %Airtime of 66.3/9.7/52.3/4.3 (total/busy/RX/TX) . Why is it RXing 52 % while there are no clients, no meshing APs? Anybody a clue?
The Noise Floor is -90, with quite some PHY Errors.

Photo of Primož Marinšek

Primož Marinšek, AlphaDog

  • 413 Posts
  • 48 Reply Likes
Well you have no clients currently, but the history shows the AP has transferred 74MB in the 5GHZ and 179MB in the 2,4GHz bands and the calculations are done on those data.

I'd say the AP is in good condition. On the 2,4GHz you've got only 5,5% of retries and no drops, the noise floor is OK and AT looks good.
Photo of Jelle Alten

Jelle Alten

  • 60 Posts
  • 12 Reply Likes
Thanks for your reply. It seems that the values of the AirTime are pinned to the last values they had when there was a client. Or at least the value doesn't go down very fast.
Either it is an average over a longer time interval than I expected, or it is registered the same way the RF Pollution is: when there is no traffic anymore, it will keep the last known value since no new values can be registered without traffic.

Can anybody confirm this?

Here are screen shots of last night (all left the building) and this morning (nobody in yet). I confirmed that there were no stations by checking the event logs of the AP. What I find interesting is that the AirTime is still pretty high, even RX is 42% after a silent night. And the fairies obviously downloaded 5 MB during the night, while uploading 800 MB, bypassing the event logs. ;-) That might also explain the airtime...

Photo of Abed AL-R B

Abed AL-R B

  • 2 Posts
  • 0 Reply Likes
hey,

how could I troubleshoot the noise floor ?


Photo of John D

John D, AlphaDog

  • 497 Posts
  • 136 Reply Likes
That looks fantastic. Your noise floors are much lower (remember that dBm's are powers of 10, so the more negative the number, the lower the noise floor). You have a moderate number of PHY errors on 2.4 and none on 5GHz, but that's fairly impressive. Airtime utilization is almost all tx/rx and very little "busy".

I live in a residential area and my numbers are worse than yours. Are you having performance issues?
Photo of Abed AL-R B

Abed AL-R B

  • 2 Posts
  • 0 Reply Likes
First, thanks for the reply.
Second, yes , I do have a performance issue .
I understand the following, If my TX power is Full and the noise is -118 , is that a good statistics ? 'casue i thought the noise should be between -85 <-> -110 .
I do have a lot of packet lost .
Photo of John D

John D, AlphaDog

  • 497 Posts
  • 136 Reply Likes
No, I think you might be misinterpreting the original post. The noise floor, loosely speaking, represents the minimum signal strength that the AP could "hear" over ambient noise. So the lower this number (or rather, the bigger the negative number), the better. So a noise floor of -120 is better than -118, for example. Anything better than -100 is really really good.


I see in your case most of your clients are connected to 2.4GHz, but you've only transferred 300MB of data or so in total. It might be more useful to look at these stats with some more data transferred (e.g. more uptime or speed tests between reboots). It might also be helpful to open a support ticket, where you can attach support info taken from the ZoneDirector or from the AP. Those contain more detailed breakdown of performance metrics which may be more useful. From the glance that you showed, everything seems healthy to me.