I’m looking to solicit input and opinions on one vs the other. It seems that I can enable polling instead of contention on my network under the nstreme tab in under interface and wlan.
I’m unsure of the advantages and disadvantages of either.
Can someone break down the settings under N-streme (disable csma, enable nstreme, enable polling etc.)
I’m trying to sort out some basics before I get on to other more detailed questions.
Settings under N-streme:
enable nstreme - turns on the MikroTik propiratity feature that boosts wireless transmit speeds and throughput, must be enabled for ALL devices connected to the AP
framing policy - increases OTA throughput by grouping more than one ethernet frame into a single wireless frame (sorta along the lines of jumbo frames on ethernet)
enable polling - assigns time slots for the AP and stations connected to transmit to prevent collisions (think back to the days of hubs, before switches were around), makes the network operate like it is TDMA
disable csma - only works if polling is enabled. since CSMA/CA (carrier sense media access / collision avoidance) is no longer needed when every device on the network is assigned a specific time when it can transmit, eliminating collisions, you can also “turn off” the radio’s hardware that listens before transmitting to see if any other device is currently transmitting (and waits if it sees one). in a remote area with little to no other traffic on the channel this has almost no effect, but in urban areas that often have lots and lots of APs visible but with weak signals, your AP and client will still wait for them to complete their transmission before it trys to transmit. Enableing in this type of enviroment can produce results as much as 20% better throughput, averaging around a 10% boost acording to my early testing. however since the radio will transmit at the same time other devices are also transmitting, if the receiving side of the transmission sees the interfering signal with a strong enough signal, it can corrupt your signal and cause the radio to have to retransmit the entire wireless frame. Enableing this option in areas of high noise can actually decrease your wireless performance as a result of the wireless frame needing to be be retransmitted a large percentage of the time.
Many thanks,
So to use N-streme I have to turn it on at the su and the ap.
With framing policy if I drop a frame will it have a large affect on latency, error rate etc or do the benefits outweigh the cons do you think. My first thought is that supporting jumbo frames is a good thing especially with video streaming applications these days. Am I right?
Polling is more efficient than contention simply because it ‘organizes’ traffic in a sense.
I’ll keep these thoughts in mind but it sounds like I should simply have nstreme turned on. ( I don’t hear any downsides), use polling and find a framing policy that works well. Any pointers on framing policy?
I see the default framer limit is 3200. Is this Kb?
sorry, been a while since I checked back here.
correct on all accounts, the default of 3200 may or not be right for you, test if out and give it a try. just don’t use any policy other than best fit. I found the others don’t seem to work as well (specifically the exact size option)
One thing you may want to do is run some packet captures and determine your packet size distributuion, and make the framer limit be slightly larger than your 2 most common large packet sizes, plus your most common small packet size. Use that only as a starting point, but make changes and watch over time to see how things react. simply running a speed test is not a fully accurate measure of the framing limit changes as packet sizes can really affect the performance / throughput