Hey All,

So we've finally decided that we would take the plunge and see what we can acheive with NV2. A bit of background; we have around 100 transmission towers; well over 250AP's, and many MANY clients We've been operating for about 7 years, exclusively Mikrotik. Clients range from RB112 / 133 / 411 / 711 ... you get the idea. Distances in this test are upto 34km; channel widths vary; signal strengths vary and i will do my best to give exacts.

We have been hugely vocal with MT on these forums and via support@, giving MT staff heaps of grief about early releases and bugs ; they will know who we are I thought it was about time that we showed them that the effort they are putting into their new releases is really really really appreciated! 5.7 so far has been amazingly stable for us (and we use snmp, nstream, nv2, mpls, vpls, bgp, ospf all on one 433 in some sites) so thats saying something! MT we really do love you!!!!

Anyway, Down to the the testing.

AP's were set with NV2, 3ms, 10km

We switched 3 of our sites and all associated clients over to ROS 5.7 & NV2 from 4.16 Nstream (we didnt bother testing NV2 with previous versions).

First site had 3 clients
Second site had 6 clients
Third site has 9 clients

We do have sites with over 40 clients; we havent migrated these yet as we are still very much in testing with Nv2 and didnt want th stress! as mentioned above clients run a whole variety of hardware platforms.

First site runs Vlan back to our MPLS core
Second site has OSPF, MPLS and BGP on it
Third site runs vlans back to MPLS code

Testing was from the client to the AP directly.

First site is 433
Second site is 433AH
Third site is 333

Signals of the tested clients range from -55 to -80

First site clients went from acheiving approx 12mbit (from nstream, 10mhz channel width to NV2) to 18.5mbit! No other changes!!!
Second site clients went from acheving approx 8mbit (from nstream, 10mhz channel width to NV2) to 17mbit!!!
Third site clients went from 8mbit sold to (from nstream, 10mhz channel width to NV2) 18mbit

Latency at idle was bouncy, upto 50ms, under load, no more than 20ms (for all 3 AP's) mostly hanging around 9 to 12ms. I guess using 3ms increased the latency to this point.

I understand that Nv2 ignores locked data rates; which under nstream **ALL** clients were locked to 24mbit, A mode only. Testing was only one client at a time. Aggregate throughput when testing multiple clients remained within about 500k of the max throughput seen from one client only.

Now for the P2P test.

This link is 34km, -78 signal (not great) 20mhz channel width. Nstream gets 28mbit one way (802.11N); NV2...very very poor and bouncy between 17mbit/24mbit. So it seems that with a less than awesome signal over long distances in a P2P enviroment Nstream still wins, hands down. Again this was set to 3ms 35km.

Take from this what you will, but we are impressed. VERY impressed. We will monitor over the next 14 days and check for memory leaks or any other odd behavior and then move to upgrade our entire platform to 5.7.

Final word, WELL DONE MIKROTIK!!!!!!!!!!!!!!

Edit: this message got declined from posting due to too many smiles!

Thank you! Yes, this is why we made Nv2, we have many people with success stories like this, I hope you can tell yours at the next MUM in your area We are working on remaining issues, so Nv2 will only get better with time.

No worries.

Normis, would it be possible to get some hard facts from your engineers about the performance of NV2 when signals are -75dB or worse; and distances are greater than 16km. Also some facts around rate selection (and the effect fixing rates has on throughput overall)

When we loaded NV2 up on an AP with 41clients, signals mostly -75 or worse and distances up to 24km; we didnt see the type of increases seen above; if anything, things got slightly worse :/ AP and clients running 5.7 NV2; 3ms and set to 27km.

We havent extensively tested yet (we will) but more information surrounding NV2 in a less than optimal enviroment would be super helpful.

Thanks!

It seems clear that the desired NV2's performance has only under optimal conditions. Unfortunately it is rare for someone to have this condition: 1W EIRP means having a lot of customers close to -80.
I also believe performance is much affected by interference from nearby channels (or better: nearby radios). Mikrotik's radios are not sufficient in this sense in my opinion. In these cases I have better performance using nstreme than nv2.

Hi

Just want to add to this post. We have had great success with NV2, in fact its probably the best thing thats happend to us. We have mutiple base stations with the highest one around 60 clients in a high interfearance area. The signalls range from -65 to -85 and at 8 to 56km it is unbielieably more stabe than it was with just nstream or 802.11. All clients get better data throughput and links stay up as long as there is power at the client. I also noticed the latency doesnt shoot up like before even if the base station gets very busy.

I have one base station giving problems in NV2 on 5.7 thats data rate just hangs up and goes to 0kbps, all the clients are still connected I just cant ping them. I suspect a card or issue.

I also congradulate you mikrotik for a great add on to an already great product.

Regards

Hi

Just want to add to this post. We have had great success with NV2, in fact its probably the best thing thats happend to us. We have mutiple base stations with the highest one around 60 clients in a high interfearance area. The signalls range from -65 to -85 and at 8 to 56km it is unbielieably more stabe than it was with just nstream or 802.11. All clients get better data throughput and links stay up as long as there is power at the client. I also noticed the latency doesnt shoot up like before even if the base station gets very busy.

I have one base station giving problems in NV2 on 5.7 thats data rate just hangs up and goes to 0kbps, all the clients are still connected I just cant ping them. I suspect a card or issue.

I also congradulate you mikrotik for a great add on to an already great product.

Regards

Hi RAHQGideon,
what radio cards do you use? Are you using N technology? Wich channel?

It seems clear that the desired NV2's performance has only under optimal conditions. Unfortunately it is rare for someone to have this condition: 1W EIRP means having a lot of customers close to -80.
I also believe performance is much affected by interference from nearby channels (or better: nearby radios). Mikrotik's radios are not sufficient in this sense in my opinion. In these cases I have better performance using nstreme than nv2.

Maybe use better antenna's at the client? Use high gain antenna's on both ends and tune the radio power down so you don't step outside EIRP limits.

The main misconception amongst many users is that power solves all but its better to do some good 'listening'.
And by using high gain antenna (usually directional for client) a second issue is reduced as well; interference.

I think the observations being made in this thread are largely from people who already have highly tuned networks running under nstream; looking to take advantage of the benefits of NV2 and finding that under the same conditions which Nstream thrives, NV2 does not.

Thats my experience so far atleast

I think the observations being made in this thread are largely from people who already have highly tuned networks running under nstream; looking to take advantage of the benefits of NV2 and finding that under the same conditions which Nstream thrives, NV2 does not.

Thats my experience so far atleast

Obviousely this vary's from network to network. I had hisites where not even nstreme would stay connected for longer than 10min at a time, this was caused mostly by interferance. We changed nothing just enabled nv2 and these links are sorted. One backhaul link didnt want to run in nv2 only nstreme and still does today

I am not saying that this is the case in all enviroments just that there is a clear improvent over normal nstreme in my case.

@RAHQGideon
what radio cards do you use? Are you using N technology? Wich channel?

I never used nstreme because I found 802.11 with RTS/CTS giving me better performances.
Now after most NV2 issues are solved I like this even more. Although interferences are something you'd really need to look out for I can do now more with my AP-networks than ever before.
It performs so well I hardly feel the need to migrate my network to 802.11n which I thought was going to be the successor of legacy 802.11.

I think NV2 works like a charm under most conditions, not in the least due the enormous input this forum gave to the MT development department.

I'm forced to use nstreme when I have too many antennas in a single pole or more radio card than one in a single RB.
I haven't issue with AP single radio on curch towers: one or two antennas per archway work well with nv2. In those places there are no problems of interference from nearby radio. Coincidence?

I have one base station giving problems in NV2 on 5.7 thats data rate just hangs up and goes to 0kbps, all the clients are still connected I just cant ping them. I suspect a card or issue.

Same experience for me.

RB433AH + R52HN in a 120º sectorial.
About 15 customers between -65 and -75db (1 to 3 km max), all of them upgraded to 5.6 version:

1.- The minimum ping rises from 1ms to 12-25 ms for ALL the clients, also with nearly zero traffic in the antenna.
2.- Few hours later, the system hangs up only for certains clients: some keep working and the others loose the connection for ever. No ping, no mac-telnet, ... and they appears to be connected in the Registration tab. I only use RB411+R52 and RB711-5AHn, and there is no rule for the disconnected ones, seems to be random.
3.- When the few customers with much better signal (-56 and -58) starts to uses their 100% bandwith (2Mb in all of my customers), performance drops and pings gets much higher than using the 811 or Nstreme protocols.

So we discard the nv2 after this experience.

I have one base station giving problems in NV2 on 5.7 thats data rate just hangs up and goes to 0kbps, all the clients are still connected I just cant ping them. I suspect a card or issue.

Same experience for me.

RB433AH + R52HN in a 120º sectorial.
About 15 customers between -65 and -75db (1 to 3 km max), all of them upgraded to 5.6 version:

1.- The minimum ping rises from 1ms to 12-25 ms for ALL the clients, also with nearly zero traffic in the antenna.
2.- Few hours later, the system hangs up only for certains clients: some keep working and the others loose the connection for ever. No ping, no mac-telnet, ... and they appears to be connected in the Registration tab. I only use RB411+R52 and RB711-5AHn, and there is no rule for the disconnected ones, seems to be random.
3.- When the few customers with much better signal (-56 and -58) starts to uses their 100% bandwith (2Mb in all of my customers), performance drops and pings gets much higher than using the 811 or Nstreme protocols.

So we discard the nv2 after this experience.

I'd question if this is due the use of NV2 solely. Maybe some configuration combination or ill configuration in some other part of ROS is causing this.
To both of you I would suggest to sent an e-mail to support so they can take a look.
In general NV2 works for many of us like charm.

Thank you! Yes, this is why we made Nv2, we have many people with success stories like this, I hope you can tell yours at the next MUM in your area We are working on remaining issues, so Nv2 will only get better with time.

Loved NV2 and R52Hn dual pol sectors at 5 Ghz clients are mostly SXT and 711 all ROS 5.7.
Charming stable network.

Thank you! Yes, this is why we made Nv2, we have many people with success stories like this, I hope you can tell yours at the next MUM in your area We are working on remaining issues, so Nv2 will only get better with time.

Loved NV2 and R52Hn dual pol sectors at 5 Ghz clients are mostly SXT and 711 all ROS 5.7.
Charming stable network.

mahnet,

can you tell s.th. on your configuration:

- Channel Width
- Signals
- CPEs per Sector ...

Channel width 20MHz
Signal -66/70 for distance 4km on SXT
CPE on one sector is 9-13
Bandwidth drawn by each sector is almost 6 Mbps & I am sure it can handle a couple more.

@RAHQGideon
what radio cards do you use? Are you using N technology? Wich channel?

R5H and R52Hn
5260 to 5825
Not running N on sectors only backhauls

I have one base station giving problems in NV2 on 5.7 thats data rate just hangs up and goes to 0kbps, all the clients are still connected I just cant ping them. I suspect a card or issue.

Same experience for me.

RB433AH + R52HN in a 120º sectorial.
About 15 customers between -65 and -75db (1 to 3 km max), all of them upgraded to 5.6 version:

1.- The minimum ping rises from 1ms to 12-25 ms for ALL the clients, also with nearly zero traffic in the antenna.
2.- Few hours later, the system hangs up only for certains clients: some keep working and the others loose the connection for ever. No ping, no mac-telnet, ... and they appears to be connected in the Registration tab. I only use RB411+R52 and RB711-5AHn, and there is no rule for the disconnected ones, seems to be random.
3.- When the few customers with much better signal (-56 and -58) starts to uses their 100% bandwith (2Mb in all of my customers), performance drops and pings gets much higher than using the 811 or Nstreme protocols.

So we discard the nv2 after this experience.

I soted this out, seems like it was the card. We changed it and all is ok now