WOW

RouterOS General
1

What’s new in v2.8.14:

*) added nstreme protocol to wireless;

That was a goooood surprise :slight_smile:

2

Unfortunetly, the huge packet loss on AR5211 is still here :frowning: Tomorrow I’ll test it on AR5212

3

hmmmmm another link with AR5211 working on 5210 turbo mode has better performance and no packet loss!!! I don’t know what’s going on here… Can you give some more info about changing in n-streme?? Did you use more frequencies?

4

This link now has a throughput of 44 MBit (from 22) at tcp bandwith-test… I saw the station-wds and wds-slave modes… I’ve made a link using station-wds, but what’s the purpose of wds-slave?

5

Sorry for posting agasin, but I put that “working” link at 5760 and it become very unstable like the other… I have better signal strenght then at 5210 :frowning:

6

There is also more CPU usage… About double times more (from 20% to 40% average)

7

With wireless-test-2.8.15 everiting works great :slight_smile:

8

hi ASM,

what does it mean that with 2.8.15-test EVERYTHING works great?

No packet loss? Highest throughput? No strange / weird behaviour ?

I’m not willing to upgrade my backbone links to find out some other annoyances, please post your detailed observations. I thought 2.8.14-nstreme and 2.8.15-nstreme were identical… probably they are not (?)

thnx, mp3turbo.

9

no packet loss, good packet latency, normal CPU usage and high bandwidth (about 25 MBps at 7 km in 5ghz-turbo)… but now I’ve updated to 2.8.16 and the results are this:

  1. With new wireless packet in 5ghz-turbo I have lower timings in the same link:
1000 packets transmitted, 1000 packets received, 0% packet loss
round-trip min/avg/max = 1/1.2/4 ms

packet size is 1500 byte

/tool bandwidth-test xxx.yyy.zzz.www direction=receive  protocol=tcp random-data=no
 size=1500
                  status: running
                duration: 4s
              rx-current: 29.6Mbps
    rx-10-second-average: 29.2Mbps
        rx-total-average: 29.2Mbps

better throughput

/tool bandwidth-test xxx.yyy.zzz.www direction=both  protocol=tcp random-data=no si
ze=1500       
                  status: running
                duration: 5s
              tx-current: 18.1Mbps
    tx-10-second-average: 17.7Mbps
        tx-total-average: 17.7Mbps
              rx-current: 18.1Mbps
    rx-10-second-average: 17.7Mbps
        rx-total-average: 17.7Mbps

same here


Starting n-streme: framer-policy=exact-size, framer-limit=3200

1000 packets transmitted, 969 packets received, 3% packet loss
round-trip min/avg/max = 1/3.2/41 ms

again at 1500 bytes

                  status: running
                duration: 13s
              rx-current: 29.8Mbps
    rx-10-second-average: 29.5Mbps
        rx-total-average: 29.5Mbps

same throughput???

                  status: running
                duration: 10s
              tx-current: 17.8Mbps
    tx-10-second-average: 17.8Mbps
        tx-total-average: 17.8Mbps
              rx-current: 17.9Mbps
    rx-10-second-average: 17.9Mbps
        rx-total-average: 17.9Mbps

yes, I think so!

Hardware used:

Point A: Celeron @ 2.6 GHz, Motherboard Intel D865PERL, Atheros 5211
Point B: IBM box, Pentium II @ 400 MHz, Atheros 5211

Edit: Add 6 Mbits traffic going all the time :slight_smile:

10

did you test from the router to the router ? that’s not recommended because the router has to generate the traffic and it takes up resources. you should test over the link. like from A1 to B1 in A1-A-B-B1