performance decrease in rb333 v3.0?

promind · Tue Jan 22, 2008 9:49 am

Hello MikroTik!
I've just purchased 2 rb333 with built-in ROS 3.0rc6.
Bandwidth test were not exactly satisfying but I can handle that/45Mbps single direction MAX/.

Now the tricky part...
I've upgrated ROS to v3.0 stable and what happened?!? 25Mbps MAX /single direction/.
Isn't it supposed to gain higher bandwidth when releasing new version instead of decreasing it?
When shall this be fixed so I can be able to switch to stable version of ROS?
P.S:I've gained better bandwidth with 2x333Mhz P2 x86.

As Apple Inc. made their commercial several years ago:

The PowerPC G3 just BURNS the Pentium 2 Processor.

Is it really?

Tue Jan 22, 2008 9:50 am

Maybe post some details about your config and testing methods? A supout.rif file to support would be even better

promind · Tue Jan 22, 2008 10:01 am

Maybe post some details about your config and testing methods? A supout.rif file to support would be even better

Sorry to hear that you don't even know about that...
Yesterday I've spoken with a friend of mine who also got a pair of 333's and he made the same mistake./upgraded to 3.0rc9/. Same problem with him!
My config is the simpliest to transparent bridge 2 networks.
RB1 - bridge , RB2 - station wds
bridge1 with NONE STP/RSTP decreases 3Mbps/
ip_conntrack switched off.
firewall OFF.
FTP/TELNET OFF.

2 AR5413 miniPCI cards / one per board / sorry for my mistake cards are XR5 which are AR5413 not 5416/which is 802.11n/

Tue Jan 22, 2008 10:17 am

a supout.rif that is made during the max traffic would be very welcome.

promind · Tue Jan 22, 2008 10:23 am

a supout.rif that is made during the max traffic would be very welcome.

Sent email/ not when max traffic/
I'll resend you during max traffic.

a-l-i-e-n · Tue Jan 22, 2008 1:28 pm

Since I've upgraded to 3.0 my clients get a lot of ping time out...
AP is RB 333 and clients are 2XRB411,1XRB112
AP and RB411clients have 3.0 installed
RB112 has MT:2.9.45 installd <--- with no ping time out

at first i thought that iterference cause this, but I tryed evry allmoust every channel on 2.4 GHz

what is wrong with 3.0 version?

wirdo · Tue Jan 22, 2008 5:34 pm

second station at -85 is your problem. Remove it and everything should be fine

a-l-i-e-n · Tue Jan 22, 2008 11:35 pm

it worked fine befor upgradeing it to 3.0... and if I switch channal signal increases to -77 dBm with the same problem (ping time out)

jwcn · Wed Jan 23, 2008 2:50 am

Try shutting the -85/-77 customer off and see what happens.

Also, try decreasing your power.

Alex · Wed Jan 23, 2008 6:02 pm

as i wroted somewhere here, with 333 and rc10 i`ve got around 55 Mbps real bandwitch.config was 5-turbo, nstreme, exact size, w/o connection tracking,wds bridge.after that i uprgaded link to 3.0 and got only 40 mbps. after that i downgraded to rc10 and all works fine. i can`t do supout because it is real link and I can`t do such expirements.

Thu Jan 24, 2008 9:45 am

how is "supout.rif" a link-threatening experiment? it's just configuration output file

promind · Thu Jan 24, 2008 11:02 am

how is "supout.rif" a link-threatening experiment? it's just configuration output file

To Alex:
I'm really amazed how you managed to get 55Mbps...
I switched off firewall disabled connection tracking and also disabled ftp/telnet
the bridge isn't stp/rstp and of course used wds not EoIP and the maximum throughoutput which I managed to get is 45-47Mbps.
After upgrade to ROS v3.0 stable I got only 25-27Mbps almost 2 times decrease in speed.
Can you share your config file or just explain to us how you did 55Mbps throughoutput?
P.S: Ether2Ether test also got 47Mbps MAX.
P.S part 2: And why MikroTik allows OFFTOPIC posts in my post?

Alex · Thu Jan 24, 2008 2:09 pm

Config from AP


/interface wireless 
set 0 ack-timeout=dynamic adaptive-noise-immunity=yes allow-sharedkey=no \
    antenna-gain=0 antenna-mode=ant-a area="" arp=enabled band=5ghz-turbo \
    basic-rates-a/g=6Mbps basic-rates-b=1Mbps burst-time=disabled comment="" \
    compression=no country=no_country_set default-ap-tx-limit=0 \
    default-authentication=yes default-client-tx-limit=0 \
    default-forwarding=yes dfs-mode=none disable-running-check=no disabled=no \
    disconnect-timeout=3s frame-lifetime=0 frequency=5300 \
    frequency-mode=superchannel hide-ssid=no hw-retries=4 \
    mac-address=00:80:48:4A:F1:51 max-station-count=2007 mode=bridge mtu=1500 \
    name="wlan1" noise-floor-threshold=default on-fail-retry-time=100ms \
    periodic-calibration=default periodic-calibration-interval=60 \
    preamble-mode=both proprietary-extensions=post-2.9.25 \
    radio-name="0080484AF151" rate-set=default scan-list=default \
    security-profile=default ssid="gorka" \
    station-bridge-clone-mac=00:00:00:00:00:00 \
    supported-rates-a/g=6Mbps,9Mbps,12Mbps,18Mbps,24Mbps,36Mbps,48Mbps,54Mbps \
    supported-rates-b=1Mbps,2Mbps,5.5Mbps,11Mbps tx-power-mode=default \
    update-stats-interval=disabled wds-cost-range=50-150 \
    wds-default-bridge=bridge1 wds-default-cost=100 wds-ignore-ssid=no \
    wds-mode=dynamic wmm-support=disabled 
/interface wireless nstreme 
set wlan1 comment="" disable-csma=no enable-nstreme=yes enable-polling=yes \
    framer-limit=3200 framer-policy=exact-size

Config from "client"

    wpa-pre-shared-key="" wpa2-pre-shared-key="" 
/interface wireless 
set 0 ack-timeout=dynamic adaptive-noise-immunity=yes allow-sharedkey=no \
    antenna-gain=0 antenna-mode=ant-a area="" arp=enabled band=5ghz-turbo \
    basic-rates-a/g=6Mbps basic-rates-b=1Mbps burst-time=disabled comment="" \
    compression=no country=no_country_set default-ap-tx-limit=0 \
    default-authentication=yes default-client-tx-limit=0 \
    default-forwarding=yes dfs-mode=none disable-running-check=no disabled=no \
    disconnect-timeout=3s frame-lifetime=0 frequency=5300 \
    frequency-mode=superchannel hide-ssid=no hw-retries=4 \
    mac-address=00:80:48:4A:F1:55 max-station-count=2007 mode=station-wds \
    mtu=1500 name="wlan1" noise-floor-threshold=default \
    on-fail-retry-time=100ms periodic-calibration=default \
    periodic-calibration-interval=60 preamble-mode=both \
    proprietary-extensions=post-2.9.25 radio-name="0080484AF155" \
    rate-set=default scan-list=default,5300 security-profile=default \
    ssid="gorka" station-bridge-clone-mac=00:00:00:00:00:00 \
    supported-rates-a/g=6Mbps,9Mbps,12Mbps,18Mbps,24Mbps,36Mbps,48Mbps,54Mbps \
    supported-rates-b=1Mbps,2Mbps,5.5Mbps,11Mbps tx-power-mode=default \
    update-stats-interval=disabled wds-cost-range=50-150 \
    wds-default-bridge=none wds-default-cost=100 wds-ignore-ssid=no \
    wds-mode=disabled wmm-support=disabled

link status

                                                              
                 status: connected-to-ess
                   band: 5ghz-turbo
              frequency: 5300MHz
                tx-rate: "54Mbps*2"
                rx-rate: "54Mbps*2"
                   ssid: "gorka"
                  bssid: 00:80:48:4A:F1:51
             radio-name: "0080484AF151"
        signal-strength: -57dBm
     tx-signal-strength: -59dBm
            noise-floor: -94dBm
        signal-to-noise: 37dB
                 tx-ccq: 96%
                 rx-ccq: 97%
           p-throughput: 53015
         overall-tx-ccq: 96%
  authenticated-clients: 1
               wds-link: yes
                nstreme: yes
                polling: yes
          csma-disabled: no
           framing-mode: exact-size
          framing-limit: 3200

RC10, connection tracking is disabled, ftp/telnet is running. from first console I run /tool bandwitch-test address=11.11.11.131 protocol=udp direction=transmit and from second monitor-traffic wlan1

received-packets-per-second: 1527
     receive-drops-per-second: 0
     received-bits-per-second: 5.9Mbps
      sent-packets-per-second: 5564
    transmit-drops-per-second: 0
         sent-bits-per-second: 57.6Mbps
                                                                               
  received-packets-per-second: 1554
     receive-drops-per-second: 0
     received-bits-per-second: 5.8Mbps
      sent-packets-per-second: 5691
    transmit-drops-per-second: 0
         sent-bits-per-second: 59.3Mbps
                                                                               
  received-packets-per-second: 1448
     receive-drops-per-second: 0
     received-bits-per-second: 5.0Mbps
      sent-packets-per-second: 5732
    transmit-drops-per-second: 0
         sent-bits-per-second: 60.4Mbps
                                                                               
  received-packets-per-second: 1399
     receive-drops-per-second: 0
     received-bits-per-second: 4.8Mbps
      sent-packets-per-second: 5579
    transmit-drops-per-second: 0
         sent-bits-per-second: 58.7Mbps
                                                                               
  received-packets-per-second: 1486
     receive-drops-per-second: 0
     received-bits-per-second: 4.9Mbps
      sent-packets-per-second: 5797
    transmit-drops-per-second: 0
         sent-bits-per-second: 60.6Mbps
                                                                               
  received-packets-per-second: 1418
     receive-drops-per-second: 0
     received-bits-per-second: 5.0Mbps
      sent-packets-per-second: 5788
    transmit-drops-per-second: 0
         sent-bits-per-second: 61.1Mbps
                                                                               
  received-packets-per-second: 1479
     receive-drops-per-second: 0
     received-bits-per-second: 5.2Mbps
      sent-packets-per-second: 5767
    transmit-drops-per-second: 0
         sent-bits-per-second: 60.5Mbps
                                                                               
  received-packets-per-second: 1477
     receive-drops-per-second: 0
     received-bits-per-second: 5.3Mbps
      sent-packets-per-second: 5557
    transmit-drops-per-second: 0
         sent-bits-per-second: 58.1Mbps

To Normis: this link go to small town, about 100 customers. When I installed this link, rb`s was with 3.0 and Speed was slow. When I downgraded to RC10 I got Good Speed Link.I had no time to make supout.

mxena · Fri Jan 25, 2008 8:53 pm

Hello,

I have exactly the same problem. I use the following schema:

Station wds 1 <---> Central Ap (2 wireless) <--> Station wds 2

Each station has a R333 with R52H and the Central Ap has two R52H. I think the problem has something to do with the bridge in the Central Ap.

When I perform a bandwith test Station 1 to Central Ap I have a througput up to 25 Mbps. The same from Station 2 to Central Ap. When I try a bandwith test from Station 1 to Station 2 I have just 10Mbps.

Connection tracking is disabled and so on. I attach a supout.rif from the Central Ap during the bandwitch test. I hope it should be usefull.

Tranks,

Miquel.

karyal · Mon Jan 28, 2008 1:04 am

Hi, just my two cents... one of our main links is running up to 60Mbps.
Signal is not good enough to reach solid 54Mbs*2 negotiation, and is running rc6
Bandwith test on this link (tcp) reported speed is 42Mbps one way.. still when in production it runs from 55 to 60 Mbps (it's "cut" to 55 to have some fading margin).
I upgraded several units to 3.1 (no 3.0) and so far i cannot see any difference in speed:

this is a unit running bandwith test + normal traffic:

received-packets-per-second: 2412
receive-drops-per-second: 0
received-bits-per-second: 6.6Mbps
sent-packets-per-second: 5334
transmit-drops-per-second: 0
sent-bits-per-second: 38.1Mbps

received-packets-per-second: 2490
receive-drops-per-second: 0
received-bits-per-second: 6.7Mbps
sent-packets-per-second: 5378
transmit-drops-per-second: 0
sent-bits-per-second: 37.9Mbps

signal is -50

you can see it gets 42Mbps/sec half duplex (as i would expect during bandwith-test)
Did you try to upgrade to 3.1 (i really made no tests with 3.0)

promind · Mon Jan 28, 2008 10:13 am

Hi, just my two cents... one of our main links is running up to 60Mbps.
Signal is not good enough to reach solid 54Mbs*2 negotiation, and is running rc6
Bandwith test on this link (tcp) reported speed is 42Mbps one way.. still when in production it runs from 55 to 60 Mbps (it's "cut" to 55 to have some fading margin).
I upgraded several units to 3.1 (no 3.0) and so far i cannot see any difference in speed:

this is a unit running bandwith test + normal traffic:

received-packets-per-second: 2412
receive-drops-per-second: 0
received-bits-per-second: 6.6Mbps
sent-packets-per-second: 5334
transmit-drops-per-second: 0
sent-bits-per-second: 38.1Mbps

received-packets-per-second: 2490
receive-drops-per-second: 0
received-bits-per-second: 6.7Mbps
sent-packets-per-second: 5378
transmit-drops-per-second: 0
sent-bits-per-second: 37.9Mbps

signal is -50

you can see it gets 42Mbps/sec half duplex (as i would expect during bandwith-test)
Did you try to upgrade to 3.1 (i really made no tests with 3.0)

My boards are rev.1 so ..one of the "old ones" as the seller said.
I've got upto 25 bwtest and in real test not much than 33/34Mbps single connetion..I just can't imagine if customers get using torrents what will happen...
3.1 has no fixed pci bus speed as MikroTik support said..I should be waiting for 3.2 where this issue will be resolved.
I apologize for my not so good English but I've been practising it just on similar forums.

karyal · Mon Jan 28, 2008 11:21 am

My boards are rev.1 so ..one of the "old ones" as the seller said.

The "55mpbs" board i was mentioning are probably the first rb333 i got..
i'm not sure what revision are, but are surely one of the "old" ones

I've got upto 25 bwtest and in real test not much than 33/34Mbps single connetion..I just can't imagine if customers get using torrents what will happen...

Are you using turbo mode? if yes, you should get much more.. if not, 34Mbps it's pretty much all you can squeeze from a single channel...

Bye,
ricky

promind · Tue Jan 29, 2008 12:46 pm

My boards are rev.1 so ..one of the "old ones" as the seller said.
The "55mpbs" board i was mentioning are probably the first rb333 i got..
i'm not sure what revision are, but are surely one of the "old" ones

I've got upto 25 bwtest and in real test not much than 33/34Mbps single connetion..I just can't imagine if customers get using torrents what will happen...
Are you using turbo mode? if yes, you should get much more.. if not, 34Mbps it's pretty much all you can squeeze from a single channel...

Bye,
ricky

Yes I use 40Mhz channels/a.k.a turbo

/
I've got approximately 48Mbps and that by my vision is not just pretty much..it's better than I expected.

karyal · Sat Feb 02, 2008 8:50 pm

Yes I use 40Mhz channels/a.k.a turbo /
I've got approximately 48Mbps and that by my vision is not just pretty much..it's better than I expected.

MT does a good job indeed... the best performance i was able to get (with 2.9) with enough cpu, nstreme , turbo, single radio has been 82Mbit/sec UDP - 72 Mbit/sec TCP - half duplex
Which was pretty impressive IMHO.
I should be able to make some test with 3.0 and RB600 next week.
Of course speed is never enough, and one the most appreciated things would be reaching such speeds with less radio spectrum usage...

Great Man · Mon Feb 11, 2008 4:55 am

RB333, RC6, Atheros 5213, 5GHz-Turbo, N-Streme On, Polling On, Exact size 3200, 54 Mb, Signal -63/-62, StN 33, CCQ 100/100, Firewall Off, Conn Tracking Off:

rx-current: 74.3Mbps
rx-10-second-average: 74.0Mbps
rx-total-average: 56.6Mbps
lost-packets: 655
random-data: no
direction: receive
rx-size: 1500

karyal · Mon Feb 11, 2008 10:47 pm

uhm... i just upgraded the main link to 3.2 (hardware is completely the same)
I did not have any performance hit (i.e. throughput), but i had a huge latency increase (from 20ms to 250ms)
I tried to tune up the link, but nothing changed.. i reverted one of the rb33 (the one running an sr5, second one is running a CM9) to rc6 and latency issue disappeared..
any idea?

Tue Feb 12, 2008 1:04 pm

uhm... i just upgraded the main link to 3.2 (hardware is completely the same)
I did not have any performance hit (i.e. throughput), but i had a huge latency increase (from 20ms to 250ms)
I tried to tune up the link, but nothing changed.. i reverted one of the rb33 (the one running an sr5, second one is running a CM9) to rc6 and latency issue disappeared..
any idea?

Could you provide us with the support output file from the router when you have latency issues? Send it to support@mikrotik.com
Are you using the same hardware in both cases - just the software version is different?

karyal · Tue Feb 12, 2008 1:11 pm

uhm... i just upgraded the main link to 3.2 (hardware is completely the same)
I did not have any performance hit (i.e. throughput), but i had a huge latency increase (from 20ms to 250ms)
I tried to tune up the link, but nothing changed.. i reverted one of the rb33 (the one running an sr5, second one is running a CM9) to rc6 and latency issue disappeared..
any idea?
Could you provide us with the support output file from the router when you have latency issues? Send it to support@mikrotik.com
Are you using the same hardware in both cases - just the software version is different?

I can re-upgrade the router to reproduce the issue..
Yes hardware is exactly the same

Tue Feb 12, 2008 1:37 pm

uhm... i just upgraded the main link to 3.2 (hardware is completely the same)
I did not have any performance hit (i.e. throughput), but i had a huge latency increase (from 20ms to 250ms)
I tried to tune up the link, but nothing changed.. i reverted one of the rb33 (the one running an sr5, second one is running a CM9) to rc6 and latency issue disappeared..
any idea?
Could you provide us with the support output file from the router when you have latency issues? Send it to support@mikrotik.com
Are you using the same hardware in both cases - just the software version is different?
I can re-upgrade the router to reproduce the issue..
Yes hardware is exactly the same

you can make the support output file from both versions so we could compare.

karyal · Fri Feb 15, 2008 5:44 pm

Uldis, i should be able to create the supout today.
It's no problem to create the supout for the "working" situation, but to create the second one i have to slow down one of the main backbones.
Sorry for the delay,
Ricky

sistelec · Mon Mar 17, 2008 10:05 pm

Hello Normis
what I have running with ros 2.9x works fine, but what I have is the same problem with ros 3x pines ill, because we do not give a compilation of version 2.9x for rb333 and rb600.

Greetings

Tue Mar 18, 2008 10:16 am

Hello Normis
what I have running with ros 2.9x works fine, but what I have is the same problem with ros 3x pines ill, because we do not give a compilation of version 2.9x for rb333 and rb600.

Greetings

there will never be a 2.9 version for new routerboards, so we have to solve your issues in v3. make full description of the problem, and send us the supout.rif files.

complete2006 · Thu Mar 20, 2008 11:14 am

I can confirm the performance decrease ...
The max I see with 5G-Turbo ist 26 MBit TCP in one direction. The RB333 are coupled back to back with an 60db attentuation. Signal ist 60 db. Wireless bandwidth is nailed to 54 MBit (*2)
I checked it with routing, wds and eoip. Always conntrack off. Traffic generator are two pc and not the routerboards...

Who has more?

promind · Sat Mar 22, 2008 12:32 am

I can confirm the performance decrease ...
The max I see with 5G-Turbo ist 26 MBit TCP in one direction. The RB333 are coupled back to back with an 60db attentuation. Signal ist 60 db. Wireless bandwidth is nailed to 54 MBit (*2)
I checked it with routing, wds and eoip. Always conntrack off. Traffic generator are two pc and not the routerboards...

Who has more?

The bandwidth test eats all your board's cpu.
Try "dandwidth" test.

nowires · Sat Mar 22, 2008 9:34 pm

Like I have commented in other thread, RB333 with RouterOS 3.4 gives me 70mbps (Turbo A)

blue · Fri Mar 28, 2008 4:32 pm

Have the same problem with 3.6. We have tested on two rb600,two CM9, 5GHz-TURBO, nstreme.... and gain 70mbit/s in UDP bandwith and ONLY 30mbit/s in TCP. max transfer between two notebooks never excedded 3-4MB/s. Few weeks ago when I was doing the same test with first RB600 in country transfer was arround 7MB/s. I realy dont remember what version of ROS was on that RB600...

Is there a solution, or just send us the link for downloading the older versions...

Thanx.....

karyal · Sat Mar 29, 2008 7:47 pm

Is there a solution, or just send us the link for downloading the older versions...
Thanx.....

Hi, about my message, i can confirm that after trying with ROS 3.6 (on RB333) the link performance increased.. they're not still at the same level of rc6, but at least i can get a 47/48 mbps half duplex consistent, and with no problem, except for latency rising at 20/30 ms because of the board cpu load.
About rb 600, our test showed that with optimal conditions (turbo, snr more than 40, consistent 54*2 negotation, 100% tx and rx ccq), we can reach 82/83 mbps UDP and 72 TCP, half duplex bandwith test.
We tested the board by using two boards for the link, and four rb 600, on the ethernet, as traffic generators.
One rb600 was not enough to generate enough traffic (max was around 50mpbs).
Note that when we run the bandwith test from the rb600 that were part of the link, fdx test was HORRIBLE (around 10mbps rx and 3 mbps tx), so don't count on that

use something different to test the link, like two rb, or a good pc

bye,
ricky

RK · Fri May 02, 2008 11:33 am

Is this fixed in 3.7? Anyone tried it?

performance decrease in rb333 v3.0?

performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?

Re: performance decrease in rb333 v3.0?