ACK window statistics how to?

RouterOS Wireless Networking
1

We are looking for some way to grab statistics about the ACK window on an MT access point, but dunno how to. Please, is there someone that could help us?

I would like to measure packets re-transmitted vs throughput to better understand/manage a overall AP quality index, if possible. Any suggestions to acomplish a way to measure that AP quality on the wireless media, using snmp or not, are welcome.

tia
Ozelo

2

Nothing? sighs

I have notice that we can grap from a wireless interfaces some statistics like out unicast packets, out octets and out errors. I didnt know exactly what are these errors, but they are a lot on wireless. We have only one wireless link that show zero errors and I wonder if this card is not working properly instead believe that there are really no errors. :slight_smile:

Perhaps I could do something with the numbers mentioned above, please advise. I was trying to do the following:

((out unicast pckts)/(out errors))/uptime = some quality index OR

((out unicast pckts)/(uptime))/(out errors) = some quality index ETC

But didnt got a satisfatory index and barely could compare to some station statistics about two numbers: “frames tx success without retries” and “frames tx success with retries” that in most cases seems very close each other. This is telling me that 50% of the frames are being retransmitted, right? But dunno why yet… Instead of twekling until reach a better performance, I would like to have a better “statistics” or at least, better understand the statistics that we can grab from MT ROS and Dude. Or any other way to detect if an AP is going well or not, before it get really worse and customers start complaim, IF that is even possible with MT.

Also, I see that an AP with 4 clients can get about 4-5 mbits throughput total, but 70 clients barely reach 2 mbits. Guess cuz the overhead of 70 clients, but I would like to measure this, if possible too. There is a lot of other wireless variables, I know. Also, didnt understand why, each time a client disassociates with extensive data loss, the other clients get a ping jump from 10ms to 2000ms. Anyone that could help me here?

Thanks in advance.