two rb 532 and 802.11a superange performance

RouterOS Wireless Networking
1

hello,

could anyone write some tips for configure two router os on rb 532 and 802.11a without nstream and with best possible encryption.rb 532 speed is 333 and speed with in two rb532 is now only about 20Mbps.
Is WDS ok? any other bridge mode? Must use ip routing for better performance?
any hints…?

anyone know how to configure to get this results:

http://www.mikrotik.com/Documentation/wireless_test.pdf


thank you,
tomas

2

I agree that MikroTik .PDF is misleading to say the least. My testing has largely only related to NStreme2 and while total aggregate throughput is close to what the .PDF shows you will not see anything more than 20-22Mbps FDX with a RB532. The board just doesn’thave the CPU power to manage more than that with the current NStreme2 code.

Oh, and to achieve 20-22Mbps FDX with NStreme2 on RB532 boards you will need to run v2.9rc9. Later OS versions included the current v2.9.4 have a problem with the NStreme2 interface just stop passing data under high loads requiring the interface to be disabled and then re-enabled.

Best,

Brad

3

I don’t have any trouble getting 37Mbps using NStreme1 with RB532s and 2.9.4 which is what is advertised in the PDF.

I am using SR5 radio cards that have a -60dBm signal at 30km.

Ken

4

There is no problem to transmit more that 20Mbps FDX with Nstreme 2 running v2.9.4 (nonnection tracking disabled).We have 3km link:

[uldis@NSRB5PMI] interface> /sys reso print
                   uptime: 14m3s
              free-memory: 18636kB
             total-memory: 30384kB
                    model: "RouterBOARD 500"
                      cpu: "MIPS 4Kc V0.10"
            cpu-frequency: 333MHz
                 cpu-load: 100
           free-hdd-space: 35448kB
          total-hdd-space: 61440kB
  write-sect-since-reboot: 0
         write-sect-total: 0
[uldis@NSRB5PMI] interface> wireless nstreme-dual p
Flags: X - disabled, R - running
 0  R name="nstreme1" mtu=1500 mac-address=00:0B:6B:35:E4:31 arp=enabled
      disable-running-check=no tx-radio=wlan1 rx-radio=wlan2
      remote-mac=00:0B:6B:33:1B:61 tx-band=5ghz-turbo tx-frequency=5700
      rx-band=5ghz-turbo rx-frequency=5550 rates-b=1Mbps,2Mbps,5.5Mbps,11Mbps
      rates-a/g=6Mbps,9Mbps,12Mbps,18Mbps,24Mbps,36Mbps,48Mbps,54Mbps
      framer-policy=none framer-limit=4000
[uldis@NSRB5PMI] interface> wireless nstreme-dual monitor nstreme1
  rx-signal-strength: -61dBm
  tx-signal-strength: -65dBm
             rx-rate: 54Mbps
             tx-rate: 54Mbps
           connected: yes
[uldis@NSRB5PMI] interface> monitor-traffic nstreme1,
  received-packets-per-second: 3789
     received-bits-per-second: 45.4Mbps
      sent-packets-per-second: 2515
         sent-bits-per-second: 30.4Mbps
5

Also note that if the framer-policy is used and you can’t transmit the data (link dies after few seconds of data transfered) you should lower the framer-limit to the size when the data is going through. Of if you can’t get good results, then don’t use the framer-policy, set it to ‘none’

6

Hello Brad,

We understand that people get different results when testing – and that is because there are a number of variables such as configuration settings, RouterOS version, wireless hardware, platform hardware, bugs, cables, interference, antennas, environment (lot of materials where signal bounce and cancel each other), and so on.

So if you don’t get the same results as us or others, you should try to address the issues of why this happens. But, don’t accuse us or others of misleading people on purpose. Life is much too short to work with people that start making accusations unreasonably. We will not assist, support, or sell to those that takes this path.

John

7

Hi uldis,
interesting result.
I’ve ordered 2 routerboards but after many post claiming low throughput of RB with Nstreme2 I’m getting little bit worried.

What type of traffic it was(UDP/TCP)?
It’s bridged or routed?
What is the avg latency you get over link when 50% loaded?

My goal is to set-up 20Mbit/full duplex TCP link with low latency ( >5ms).
Is it a dream or reality?


8

What type of traffic it was(UDP/TCP)? UDP Bandwidth test.
It’s bridged or routed? Routed

Doing Full Btest with 20mbps both direction using UDP we had approx 8ms. With TCP approx 15ms.

9

Whoaa…guess I should check this forum more often!

First, let me say there was no intention to disparage the MikroTik product and if that impression came across I apologize.

I think we are all saying the same thing here. 44Mbps is about all the RB532 can handle. Cut it up however you want, but 44Mbps is it. If it’s 21-22Mbps FDX or 45Mbps HDX the RB532 is still is only capable of about 40-45Mbps.

I would love to see an NSTreme2 configuration loaded on two RB532s passing more than 45Mbps aggregate throughput, but I just don’t see it happening. I posted several screenshots of BW test results testing from one 3GHz MT box to another 3GHz MT box across two RB532s running NStreme2. The results don’t lie unless I’ve configured the NStreme2 setup incorrectly.

The two RB532s are sitting side by side with rssi readings at a solid -60. The NStreme2 configuration is setup exactly as the MT example shows. Please provide me a NStreme2 configuration for the RB532 that will produce better than 21-21Mbps FDX payload capacity…I don’t believe one exists.

Best,

Brad

10

My goal is to set-up 20Mbit/full duplex TCP link with low latency ( >5ms).
Is it a dream or reality?

These two screenshots will show the latency you can expect with a fully loaded 20Mbps FDX RB532 link. I haven’t tried v2.9.5 yet, but plan to this weekend.


http://www.belwave.com/NStreme2_v2p9rc9_FDX_Test_From_RM1_to_RM2_at_24MB_Rate_9-20-05.jpg

http://www.belwave.com/NStreme2_v2p9p4_FDX_Test_From_RM1_to_RM2_at_24MB_Rate_Crash_9-20-05.jpg


Not only did v2.9rc9 outperform v2.9.4 by about 2MB it also did not crash like v2.9.4. Again latency was better with v2.9rc9 as well…

36MB Rate on v2.9.4 crashed so quickly I didn’t even bother taking a snapshot.

36MB Rate on v2.9rc9 did not crash, but the results were much poorer than the 18MB or 24MB rates.

BTW, rssi readings on all tests run are solid -60.

Setup:

192.168.1.70 RM Test Unit 1 3.0GHz
Ethernet

192.168.14.100 RB532 NStreme2
Wireless
192.168.14.101 RB532 NStreme2

Ethernet
192.168.1.71 RM Test Unit 2 3.0GHz


Best,

Brad

11

Apples and oranges. I am only refering to FDX links…NSTreme2. Under load a FDX link will always outperform a HDX link… this is important for VoIP applications etc…

Best,

Brad

12

[/quote]

There is no problem to transmit more that 20Mbps FDX with Nstreme 2 running v2.9.4 (nonnection tracking disabled).We have 3km link:

[uldis@NSRB5PMI] interface> /sys reso print
                   uptime: 14m3s
              free-memory: 18636kB
             total-memory: 30384kB
                    model: "RouterBOARD 500"
                      cpu: "MIPS 4Kc V0.10"
            cpu-frequency: 333MHz
                 cpu-load: 100
           free-hdd-space: 35448kB
          total-hdd-space: 61440kB
  write-sect-since-reboot: 0
         write-sect-total: 0
[uldis@NSRB5PMI] interface> wireless nstreme-dual p
Flags: X - disabled, R - running
 0  R name="nstreme1" mtu=1500 mac-address=00:0B:6B:35:E4:31 arp=enabled
      disable-running-check=no tx-radio=wlan1 rx-radio=wlan2
      remote-mac=00:0B:6B:33:1B:61 tx-band=5ghz-turbo tx-frequency=5700
      rx-band=5ghz-turbo rx-frequency=5550 rates-b=1Mbps,2Mbps,5.5Mbps,11Mbps
      rates-a/g=6Mbps,9Mbps,12Mbps,18Mbps,24Mbps,36Mbps,48Mbps,54Mbps
      framer-policy=none framer-limit=4000
[uldis@NSRB5PMI] interface> wireless nstreme-dual monitor nstreme1
  rx-signal-strength: -61dBm
  tx-signal-strength: -65dBm
             rx-rate: 54Mbps
             tx-rate: 54Mbps
           connected: yes
[uldis@NSRB5PMI] interface> monitor-traffic nstreme1,
  received-packets-per-second: 3789
     received-bits-per-second: 45.4Mbps
      sent-packets-per-second: 2515
         sent-bits-per-second: 30.4Mbps

[/quote]

Ok, now show us what the graph history looked like…my guess is it was all over the place jumping up and down. The snapshot you provided does not show any history. My test information shows latency over the link during the test and a history of solid & consistant 21-22Mbps FDX.

http://www.belwave.com/NStreme2_v2p9rc9_FDX_Test_From_RM1_to_RM2_at_24MB_Rate_9-20-05.jpg


Sure, I could set the Rates to 54MB on my test setup as well, but links that jump all over the place from 45/30Mbps to 10/5Mbps and back are of no use to us. Many tests over and over demonstrate on the RB532 a 24MB Rate & NStreme2 setup is the most solid combination.

Solid consistant BW is what is required for our clients. If your RB532 test setup can show 5, 10, 15 minutes of sustained 45Mbps/30Mbps then you have something and I want a copy!

Best,

Brad

13

Hi,
boards arrived, so I upgraded from installed 2.9.1 to 2.9.5
and the results are:

max UDP full duplex:
28.5/29Mbit/s

max TCP full duplex:
23/8
slow cpu on machines generating traffic
when changed RBs, it was the same, so I guess CPU was limiting

max TCP one-way
29Mbit/s

when passing limited traffic, FDX 25/25Mbit
latency was from 5 to 15ms


setup:
nstreme2 in bridge with ether1, freq1 5500, freq2 5700, no-turbo(it was the same with/without), CPU clock 330MHz, framing: none, signal -55, tested on desk, linked on 54Mbit


I have a question:
I played with RBs from friend and in bios, there was an option to overclock CPU to 399MHz, but on mine, there is max 330. they are from the same delivery,same price,look same
i want to try 399, it should increse the throughpt, cause the RBs CPU is limiting. any ideas how to do it??

14

Here are a few test results running v2.9.5 on a couple RB532s:

http://www.belwave.com/NStreme2_v2p9p5_FDX_Test_From_RM1_to_RM2_at_24MB_Rate_10-1-05.jpg

http://www.belwave.com/NStreme2_v2p9p5_FDX_Test_From_RM1_to_RM2_at_36MB_Rate_Crash_10-1-05.jpg

http://www.belwave.com/NStreme2_v2p9p5_FDX_Test_From_RM1_to_RM2_at_36MB_Rate_NoFramer_10-1-05.jpg

http://www.belwave.com/NStreme2_v2p9p5_FDX_Test_From_RM1_to_RM2_at_48MB_Rate_NoFramer_10-1-05.jpg

http://www.belwave.com/NStreme2_v2p9p5_FDX_Test_From_RM1_to_RM2_at_54MB_Rate_NoFramer_10-1-05.jpg

It appears MT has fixed the NStreme lockup/crash problem as long as no Framer Policy is set. This is great news as now we’ll be able to run firmware newer than v2.9rc9 on NStreme2 links!

Although the higher Data rates are not as stable as the 24MB Data Rate they are much more symmetrical than previous versions and able to squeak out a few more Mbps.

Test setup:

MT 3GHz router 192.168.1.70
ethernet
MT RB532 NStreme2 192.168.14.100
wireless
MT RB532 NStreme2 192.168.14.101
ethernet
MT 3GHz router 192.168.1.71


Best,

Brad

15

Interesting. I believe the only two options on the RB532s I’ve checked are 266MHz & 333MHz. However, I do recall reading somewhere that MT had an optional (special order?) 400MHz RB532. Possibly your friend ended up with the optional CPU?

Best,

Brad