Hi All,

I need to make running link between Routerboards that supports 5ghz-a/n/ac working in Nstreme.
Link can't be established when I use frequency between 5500 and 5700 MHz. My country is Italy and I need to work in that range.
Out of that range that link in Nstreme working ok.
If I use NV2 all frequencies are ok.
I tried RouterOS version 6.19 and 6.21.1.

Anyone can help me?

Thank you

Just use NV2, nstreme doesnt work well on N/AC interfaces. You will have slightly higher latency but more performance and more stability with high load on the link using NV2.

Why do you want nstreme btw?

I was also going to ask same question , Why do you need the old legacy nstream over NV2?

I was also going to ask same question , Why do you need the old legacy nstream over NV2?

maybe in NV2 1 TCP connection is low throughput when AP bride mode is use with routing NV2 is ok ,but in Nstreme mode when AP or CPE is in bridge mode 1 TCP connection is throughput ok more hops too sorry for me english

http://forum.mikrotik.com/viewtopic.php ... 40#p455928

I was also going to ask same question , Why do you need the old legacy nstream over NV2?

TCP over multiple hops performed much better at times nstreme was working. nv2 has better link performance
but single users do not get the tcp performance with speed tests as they did with nstreme.

Newer nv2 is better but the problem still is there.

Well, I have one AC link using NV2 and got 140Mbps throughput with 1 TCP connection.... no problem to achieve over 100Mbps on speedtest.net.. But its just one NV2 link behing fiber and 24GHz radio.

Ok you are right, I have just switched one link in between NV2 (3 in row) where I was getting only about 10Mbps with single TCP connection from NV2 to Nstreme (single chain 40MHz N) with the settings described here:

http://forum.mikrotik.com/viewtopic.php?f=7&t=83183

And now getting nearly 50Mbps with single TCP connection.. so it really seems 3 links in chain running NV2 cause low single TCP throughput (even with 10 TCP connection I was not able to achive same throughput like I do after switching to nstreme).

Try the setting described in the thread above, it seems it help in some cases.. for example just one link in the middle of 2 NV2 links..

Well, I have one AC link using NV2 and got 140Mbps throughput with 1 TCP connection.... no problem to achieve over 100Mbps on speedtest.net..

Ok you are right, I have just switched one link in between NV2 (3 in row) where I was getting only about 10Mbps with single TCP connection from NV2 to Nstreme (single chain 40MHz N) with the settings described here:

http://forum.mikrotik.com/viewtopic.php?f=7&t=83183

And now getting nearly 50Mbps with single TCP connection.. so it really seems 3 links in chain running NV2 cause the single TCP throughput (even with 10 TCP connection I was not able to achive same throughput like I do after switching to nstreme).

Try the setting described in the thread above, it seems it help in some cases.. for example just one link in the middle of 2 NV2 links..

We replace MT Backhaul links wherever possible to solve this problem.

Yea, I would do so too of course... but until we have budget for it I need to live with MT

Thanks for all replies!
But I would know if someone had same problem with Routerboards that supports AC and Nstreme only in frequency between 5500 - 5700 MHz.

Thanks

We replace MT Backhaul links wherever possible to solve this problem.

Licensed or unlicensed backhaul links?

We replace MT Backhaul links wherever possible to solve this problem.

Licensed or unlicensed backhaul links?

24GHz UL for short hops up to a mile and licensed for longer links.

Thanks for all replies!
But I would know if someone had same problem with Routerboards that supports AC and Nstreme only in frequency between 5500 - 5700 MHz.

Thanks

Well I dont see any reason why it would not work with nstreme on specific frequency, maybe there is too much noise? What is the CCQ if you switch to NV2 or 802.11? I am using N links with nstreme and they are fine on all freqs. However not sure about AC chips but I would bet there is no difference.

Same problem by me when using nv2. It is not possible to get high speeds through multiple hops over single tcp connection.
For example I have 5 nv2 ptp links in a row. Between every hop I can get >100Mbps through bw test. But when testing over all hops, the bandwidth is 10Mb.

Same problem by me when using nv2. It is not possible to get high speeds through multiple hops over single tcp connection.
For example I have 5 nv2 ptp links in a row. Between every hop I can get >100Mbps through bw test. But when testing over all hops, the bandwidth is 10Mb.

This is known long time and ignored long time. MT needs a phase of software consolidation. Instead they implement new features and accept the reduced usability of their products. So we are urged to look elsewhere. Our customers dont accept failures and low tcp bandwidth. I cant send them a feature list of ROS to hold them back looking elsewhere. I am bored talking to speedtesters.

I was also going to ask same question , Why do you need the old legacy nstream over NV2?

TCP over multiple hops performed much better at times nstreme was working. nv2 has better link performance
but single users do not get the tcp performance with speed tests as they did with nstreme.

Newer nv2 is better but the problem still is there.

I noticed this also and discovered simple queue type was the problem. for some reason fifo does not work well with NV2. I changed the default small type to sfq and all was well again

I was also going to ask same question , Why do you need the old legacy nstream over NV2?

TCP over multiple hops performed much better at times nstreme was working. nv2 has better link performance
but single users do not get the tcp performance with speed tests as they did with nstreme.

Newer nv2 is better but the problem still is there.

I noticed this also and discovered simple queue type was the problem. for some reason fifo does not work well with NV2. I changed the default small type to sfq and all was well again

Where you setting this options ? In RB SXT Interface queues/ ether1-only hardware queues / wlan1-wireless default, In Queue Types only hardware-queue is set NONE and wireless default has sfq this is default seting , what I must do what change ? thanks