Hi to All,

recently i downloaded a one page PDF file that describes RouterBoard Performance Tests (it is attached)
.

It is mentioned that the test setup was carried out with the use of Agilent N2X equipment.

Can you provide me with more information concerning the Test Setup?

For example , what ports were used at the RB400AH @680 MHz testing, what was the traffic pattern injected (i.e. constant, burst, ramp) , or was it a RFC test setup (RFC2544)?

Thank you,

Kolpano.

better go to routerboard.com and check the product you are interested in with speed test according to RFC2544

this PDF is using older RouterOS version.

Hi Janisk,

thank you for your prompt reply.

I looked into the info you provided me from the http://routerboard.com/RB493AH :

"All throughput tests done with Xena Networks specialized test system and RouterOS v5, according RFC2544, with Ethernet frame sizes 64, 512, 1518 bytes.
Each board is tested with specified number of Ethernet interfaces, to ensure optimal load on hardware."

Where can i find more specific information about the testbed topology used for the throughput test for RB493AH?

I am asking about these testbed details because i have some question concerning this test.

For example :

1. What ports were used in order to ensure optimal load on hardware (according to one of the remarks)?

2. The 199.48 Mbps for 512 Bytes packets concerns traffic that was received (for example) from Eth1, then transmitted from Eth1 to Eth2 , received from Eth2 and then transmitted through a looping device back to Eth1 and all of this traffic added in order to provide this number of 199.48 , that is 50 Mbps per port and per direction?

Thanks again for your support,

Kolpano.

RB493 has 2 witch chips and tests where running from one port on switch1 to another port on switch2.

Data throughput means that 2 ports where connected and port1->port2 passed 50% of that value and port2->port1 sent another stream covering other 50% of the bandwidth.

Thank you Janisk,

i recall that the RB493AH has one port (the POE one) which is independent from the others. The other 8 ports (ETH2 - ETH9) are part of the same chip (ICPlus178C).

I need to do some L3 routing tests , what ports do you suggest that i should use? For the time being i randomly chose port 2 and port 3 which i suppose that they are part of same chip.

Thank you,

Kolpano.

if possible use ports that are not from the same switch chip so you are not limited by connection odswitch chip to CPU.

Hi Janisk,

i will use the POE and Eth2 and i will do again the RFC2544 tests.

I will let you know about the results.

Thanks.

Hi again,

i used the following topology for my RFC2544 testing:

TREND-UNIPRO-Analyser-Source<--------->Port-1-POE---RB493AH---Port-3<----------->TREND-UNIPRO-Analyser-Loopback-Device

The test was based on L3 routing with the Source Analyser being in a different IP Subnet from the Loopback Analyser . So the RB493AH was doing the routing for these two devices. ConnTrack was OFF.

The throughput test results for Port 1 (i.e. the source) were the following:

64 Bytes --> 7.2 Mbps
128 bytes --> 27 Mbps
256 bytes --> 46.4 Mbps
512 bytes --> 39.1 Mbps
1518 bytes --> 43.2 Mbps

Looking back in http://routerboard.com/RB493AH specs i should get 52.38 Mbps (64 Bytes frames) for routing with firewall ON and Conntrack OFF.

How was this 52.38 Mbps calculated for 64 bytes frames?

Thank you.

Hi again,

i used the following topology for my RFC2544 testing:

TREND-UNIPRO-Analyser-Source<--------->Port-1-POE---RB493AH---Port-3<----------->TREND-UNIPRO-Analyser-Loopback-Device

The test was based on L3 routing with the Source Analyser being in a different IP Subnet from the Loopback Analyser . So the RB493AH was doing the routing for these two devices. ConnTrack was OFF.

The throughput test results for Port 1 (i.e. the source) were the following:

64 Bytes --> 7.2 Mbps
128 bytes --> 27 Mbps
256 bytes --> 46.4 Mbps
512 bytes --> 39.1 Mbps
1518 bytes --> 43.2 Mbps

Looking back in http://routerboard.com/RB493AH specs i should get 52.38 Mbps (64 Bytes frames) for routing with firewall ON and Conntrack OFF.

How was this 52.38 Mbps calculated for 64 bytes frames?

Thank you.

I have no experience with that kind of testing equipment, but it is very strange that just standard FTP download trough this router can get much more than your test show.
What settings do you have on that testing equipment and what is your router config (paste /export compact)

the SWITCH 1 goes to port 2 - 9 so the performance will be 50% of the data

you must try port 1 - port (2-9) to see the real performance

Hi ,

the RB493 config is pretty straight forward.

[admin@MikroTik] > ip address print
Flags: X - disabled, I - invalid, D - dynamic
# ADDRESS NETWORK INTERFACE
0 2.2.2.1/24 2.2.2.0 ether3
1 1.1.1.1/24 1.1.1.0 ether1

[admin@MikroTik] > ip route print
Flags: X - disabled, A - active, D - dynamic, C - connect, S - static, r - rip, b - bgp, o - ospf, m - mme,
B - blackhole, U - unreachable, P - prohibit
# DST-ADDRESS PREF-SRC GATEWAY DISTANCE
0 ADS 0.0.0.0/0 192.168.1.1 0
1 ADC 1.1.1.0/24 1.1.1.1 ether1 0
2 ADC 2.2.2.0/24 2.2.2.1 ether3 0
3 ADC 192.168.1.0/24 192.168.1.10 ether9 0

Firewall is ON and Conntrack is disabled.

The Testers are configured as follows:

Tester-1-IP-1.1.1.2 <--->Port 1-493AH-Port-2 <----> Tester-2-Loop-2.2.2.2

So the tester with IP 1.1.1.2 sends traffic via RFC2544 to 2.2.2.2 which then loops back the traffic to the 1.1.1.2

I did new tests , as the previous test was based on Auto Negotiation for Port 1. Now it was set to 100 Mbps FD.

The results after putting duplexing ON are much better than my previous test:

64 Bytes ----> 38 Mbps
128 Bytes ---> 75.6 Mbps
512 Bytes ---> 96.2 Mbps
1518 Bytes ---> 98.7 Mbps

As Frame rate is concerned :

64 Bytes ---> 74405 fps
128 Bytes ---> 73910 fps
512 Bytes ---> 23496 fps
1518 Bytes ---> 8127 fps

ALL ABOVE TESTS ARE DONE with 10% Frame Loss threshold , which means after 10% the test fails.

If i choose Max Frame Loss 1% then the results are the following:

64 bytes ---> 0
128 Bytes ---> 65 Mbps
512 Bytes ---> 96.2 Mbps
1518 Bytes ---> 98.6 Mbps

Frame Rate:

64 Bytes ---> 0
128 Bytes ---> 63371 fps
512 bytes --> 23496 fps
1518 bytes ---> 8127 fps


+++++++++++++++++++++++++++++++++++

Finally i did one more test with 5% max Frame Loss Tolerance and the results are:

64 bytes ---> 38.1 Mbps
128 Bytes ---> 75.6 Mbps
512 Bytes ---> 96.2 Mbps
1518 Bytes ---> 98.6 Mbps

Frame Rate:

64 Bytes ---> 74405 fps
128 Bytes ---> 73910 fps
512 bytes --> 23496 fps
1518 bytes ---> 8127 fps


My question is how the 52.38 Mbps and the 102300 fps achieved (as above mentioned pdf file from routerboard.com mentions) for 64 bytes packets and in what specific port was that throughput achieved?

My results are for port 1.

These results are more like the results from the routerboard pdf with Conntrack is on (39.68 Mbps and 77500 fps) which is not true in my case as Conntrack is not enabled :

admin@MikroTik] > ip firewall export
# jan/02/1970 05:08:44 by RouterOS 5.15
# software id = SLXJ-AEI1
#
/ip firewall connection tracking
set enabled=no generic-timeout=10m icmp-timeout=10s tcp-close-timeout=10s \
tcp-close-wait-timeout=10s tcp-established-timeout=1d tcp-fin-wait-timeout=\
10s tcp-last-ack-timeout=10s tcp-syn-received-timeout=5s \
tcp-syn-sent-timeout=5s tcp-syncookie=no tcp-time-wait-timeout=10s \
udp-stream-timeout=3m udp-timeout=10s
/ip firewall service-port
set ftp disabled=no ports=21
set tftp disabled=no ports=69
set irc disabled=no ports=6667
set h323 disabled=no
set sip disabled=no ports=5060,5061 sip-direct-media=yes
set pptp disabled=no

Thanks,
Kolpano.

Hi to all,

one more remark is that as the traffic is leaving from the Traffic Generator at Port 1 and the traffic returns back to the Generator (full duplex mode) then my results for 512 bytes and 1518 bytes are identical to the values presented in the http://routerboard.com/RB493AH .

Is my assumption for the traffic coming in and going out correct ? Should it added as in the specification sheet seems that it is added (i mean added for both directions)

If the above is correct , my only problem seems to be the throughput for the 64 bytes packet as my results are better than the ones presented in routerboard site !

I have 38.1 Mbps X 2 = 76.2 Mbps which is higher from the 52.38 Mbps in the specifications and fps is 74405 X 2 = 148810 fps which is much higher than the 102300 fps in the specifications. My results are more similar to routerboard sepecs for bridging mode but as my traffic is in different L3 subnets i suppose that i am in routing mode (unless it acts like cisco L3 switching, first packet routed , subsequent packets L2 switched).

If all previous thoughts are correct then the specifications are fully verified by my tests.

Any thoughts?

Thank you all,

Kolpano

speed increase however could be explained by use of newer RouterOS version.

And if i understood you correctly that you sum packets that are transmitted by router (sent out on one of the interfaces) than that is correct.

also, if you check http://routerboard.com/RB493AH and scroll down you see the actual result on that page. PDF you are referring to is older data.

and in our test traffic is generated separately for each stream with 1% tolerance.

Janisk thank you for your answers,

1. I am using 5.15 RouterOS so it is very new.

2. If i sum the packets (going out and coming in) then i get very very close to the results that are mentioned at the http://routerboard.com/RB493AH (except the 64 bytes results for which i have questions mentioned in thread above)

3. I am not using the PDF File. I used it when started the tests now i rely only on http://routerboard.com/RB493AH document.

4. If i use tolerance of 1% the 64 Bytes packet throughput test fails. If i use tolerance of 3% for 64 bytes packet the test is OK and i get the document's Bridging results.

When i use tolerance of 2% the test for the 64 bytes packet also fails.

So the threshold for successful results is 3% and not the 1% you mention.

Kolpano.

you should sum only packets leaving the router - as they are successfully received, processed and sent out.

These are test conditions and settings that are set - so the tolerance is set to 1% in our test setup and we are generating 2 independent streams, one in each direction.

you should sum only packets leaving the router - as they are successfully received, processed and sent out.

These are test conditions and settings that are set - so the tolerance is set to 1% in our test setup and we are generating 2 independent streams, one in each direction.

Janisk,

i have only one stream that is the traffic is generated from the source, it is sent to the loopback device and then back to the source.

So what i send is what i receive. This means that i have only one stream that is going through the router and i get back the stream to the sending device.

I do not think that this is much different from your setup in which (if i understand correctly) you send traffic from A-->B and then another independent stream from B-->A.

Unfortunately one of my two devices is only loopback mode due to licence and i can not create two different streams.

Only problem is that my accepted tolerance is 3% and in your test is 1%.

Thanks,

Kolpano

Conclusion:

RB works faster with newer versions (as the routerboard.com results might have some age on them), or results are better with higher percentage of tolerance (that to me is more or less obvious).

It is possible that MT testing equipment have more precise testing iterations, together with multi-stream setup it allows to send in exact amount of packets that will not be dropped, this might be why 1% tolerance setup doesn't fail for them.

Thank you all for your answers and support.