12-12-2012 10:57 PM
Have read some of the archive of the forum pertaining to Airwave parameters example avg signal strength, signal quality etc, that might differ to Cisco, Aruba and so on.
For this case, we can assume that are devices are Aruba, but I am still confused what is the actual number reflected at the end users side and what is the recommended threshold
Can anymore help to explain the following parameters and its relative readings at users' end plus the recommened threshold?:
Average Signal Strength
*Signal Quality ( I believe that this will never go above 50, any reasons why it scale is till 100 ? )
Recently we have just deployed a new WLAN based on Aruba, but customer is complaining that some users are having dropped connections. Checked the Airwave and indeed there are a few AP having more than 10 Channel changes, noise floor ~ -90dbm, MAC/PHY ~ 60 and signal quality ~ 6 - 30.
Re: [Airwave Parameters]
12-13-2012 03:39 PM
Skipping signal strength since I typically consider signal strength and signal quality to be the same.
Signal quality is a normalized number which I have yet to see exceed 60 (I believe having it graph up to 100 was a past product management decision). In the Aruba HQ environment, we treat 50-60 as excellent, 30-50 as good, and anything lower than 30 as bad. Though many customers treat 40-60 as excellent and anything lower than 25 as bad - this really depends on the deployment.
Noise floor is the amount of interference calculated based on attenuation numbers (from AP to AP and AP to client data). This is calculated in dBm, and values > -80 are considered high noise. Other factors that can affect noise floor are environment variables like metal walls, concrete, etc.
For MAC / PHY errors, these are physical layer errors and data link (MAC) layer errors. Typically, you might see 50-200 frames of these type of errors per second per radio - this is a normal case in the Aruba HQ network. The less error frames, the better QOS for the end user. In average, a client may be pushing 1200-2000 frames/second, so we consider >1000 frames/second as the threshold for high errors. If you have clients pushing more frames/second on average, then you might set the threshold higher.
For the end user deployment you gave, the fact that ARM is adjusting channels often is a large concern. Ideally, you want the channel changes to be low, this may require the AP placement to need adjustments to prevent the channels from overlapping too much (especially if the user has N radios). Some additional things to consider: number of clients per AP, client roaming rate, other things affecting the air (microwaves, tvs, etc). Your noise floor and MAC/PHY error counts aren't that bad, yet the signal quality seems low to me. Is the client signal quality distribution a bell curve? Or is it skewed toward the 30? It more users are near the 30 side, then it wouldn't be so bad.
Senior QA Engineer - Network Services
Aruba Networks, a Hewlett Packard Enterprise Company
Re: [Airwave Parameters]
we lately have lots of users complaining about slowness and even disconnects. i've been looking at the 'daily rf health report' and output of
'sho ap debug radio-stats' command. i have a few questions about the output of the above
-is 'phy error (%)' the % of total traffic on the ethernet connection
to/from this AP? (i ask this question because if it is referring to the
ethernet connection, then what is it doing in the rf report?)
-are 'mac errors' a subset of physical errors or a separate count?
-in the output of the 'show ap debug radio-stats' command, i see
'rx drops' as high as %90 of 'rx frames'. this would seem to indicated lots
of rf or other emi interference, correct? however, when i look at
'most noise' in the report, i see -96 and -92 dbm for the 2.4 and 5ghz
-is my understanding of 'dropped' and 'retry' frames correct?
i understand 'dropped' as the count of frames which after multiple retries
is correctlty received. i understand 'retry' count as the number of times
frames were sent repeatedly due to receive errors. such frames, are
eventually either dropped or successfully received.
iam still collecting data, but it seems to me that the controller needs some help in setting xmit power. in 1 particular building we have 2 ap's on each floor, spaced ~5 yards from each other. any other suggestions?