We have deployed a new sectorized tower using Mikrotik AP with SR2 radios and we are installing Mikrotik with CM9 for CPE. We have enabled Nstreme and have seen great results up to this point. We only have 4 or 5 clients on each sector.
In reading the forums most posts about Nstreme are negatiive. I would like to hear sucess stories from people using Nstreme.
With only 4 or 5 clients I think you will see great results regardless of what setup you have. Once you get up to 20 or 30 active users, report back on whether it’s still working as well as you are seeing now
Nstreme is a faily good “protocol” … (still a bit of bugs However it will increase the latency a slight bit, but nothing worth a headache over ( about an extra 3ms. but it works wonders with a large p2mp enviroment.
An alternative however, that is if your intention is based on having the max amount of clients on a single ap. is to use compression.
Example: Without compression: 20Mbs/ at 3408 P/s : With 29Mb/s at 5687 P/s
I really wish people on this forum would be more clear and more specific when they are talking about their experiences with Nstreme. There is Nstreme with polling and there is Nstreme without polling. Just saying “I have problems with Nstreme” tells us nothing. Please specify whether you are using the polling feature or not. My suspicion is still that it is MikroTik’s polling implementation that is causing the problems you all are hearing about, not necessarily Nstreme itself.
We will only be able to prove this once someone who is having problems with Nstreme + Polling on a PtMP system takes the plunge and tries to turn just the polling off to see if the problem is alleviated, rather than turning Nstreme off completely. I still wish to try this myself…it’s just a matter of finding a good time to do it.
Point understood, but I suppose its because polling itself is not an “integral” part of the nstreme protocol and must only be treated as an elective. Nstreme=nstreme polling=option of nstreme.
Yes, I understand why people are forgetting to be specific, but it doesn’t change the fact that we are going to get nowhere in our discussions if we cannot figure out why some people have had success with Nstreme in a PtMP environment and others have had a miserable experience with Nstreme.
The only way we are going to be able to do that is by comparing notes and breaking down all of the factors and variables. Polling is one of those variables.
Thanks for the feedback, I have polling enabled, framer policy=none, framer limit=3200. I have read that most people have left the framer policy and limit set to default.
It seems Nstreme has worked well with low sub counts. What number of subs seems to be the upper limit.
I have some AP’s running 50 subs without Nstreme that work ok, not great but ok.
I am using polling, and framer policy = best fit. Did not show any signs of weakness.
Something went wrong with a r112, with r52, that keeps disconecting, at random times, from 24 hours to 2-3 days, it disconects. I’ve put a script that resets the board every 2 days, at nights, to overcome this.
Somewhat, with wpa-psk security, I can’t enable compression, on r52 cards.
Also tried other cards - atheros 5413 chipset. Did not work on same config, on neither 5 or 2.4. MT versions from 2.9.6 to 2.9.42. The same thing.
Is this an issue, or is it the way it should be ?
I can’t afford to leave the link without encryption.
I’ll try tonight with polling on or off… to see if it makes a difference. The link wich i’m testing is a wds bridge, not routed. No firewall applied on it.