I have been scratching my head, searching forums, wikis, google on this... No answer.
I have two CRS 226, both on 6.40.1 firmware up to date (original 3.22) in two separate buildings with a fiber cable between (No idea what type or length).
I have purchased 4 Mikrotik S+31DLC10D, housed in the SFP+ ports: port 1 to other CRS226, port 2 to server (2 servers with 10GB NIC)
All SFP+ ports LED show active and 10G

This is one end of the link



This is the other end



Here are the bandwith tests






Things might have gotten worse during all the testing (connectors dirty) but I have never been able to get more than 2000Mbps.
See below readings during file transfer between servers (connected to SFP+ 2 on each switch)





I would love to have some light shed on this mystery. I have tested the switches with a SMF cable but getting same results.

Thanks for your help!

Have tested an iperf beetwen both server?

Think about higher MTU

Heard about MTU settings... Could you please elaborate?
I know they can go all the way to 9204... Do I just plug any number?

Have tested an iperf beetwen both server?

Downloaded it , learning how to use it....

The CRS226 has a 400Mhz processor with a single 1Gbs connection to the switch fabric. That is what is limiting your speed tests using the local CPU.

https://i.mt.lv/routerboard/files/CRS22 ... 162143.png

What should be MTU if i use 10G sfp+ ? Does having a mtu of 1500 cause any problems on 10Gb interface ?

please post your configuration export

look like your are not switching (hardware), maybe you are bridging (software)

please post your configuration export

look like your are not switching (hardware), maybe you are bridging (software)

I hope this will be understandable

<<Provider>> ===802.3ad===(3x 1Gb interface) <<CCR1072>>> ===10Gb interface=== <<CCR1036>> after this all other networks 1Gb. so for this 10Gb interface do I need increase mtu size ?

Thanks

take a look at your cpu load at the moment of the tests

Testing over 1Gbps is hard - mainly due to the fact that most network cards are 1Gbps. To properly test you will need to powerful PC's on both sides of the link with 10Gbps network tests. Then run IPerf3 with multiple streams.

One PC is the server, so you would just run e.g iperf3 -s
And on the other you would run iperf3 -c 192.168.0.1 (IP address of the server) -p 10 (number of streams) -t 60 (time in seconds).
You cannot run a bandwidth test on the routers - the CPU is not powerful enough to generate the traffic. But will easily pass through the traffic.

Hi,
no, traffic testing is not a problem, I would like to learn that if I need to change the mtu value for 10gb interfaces ? because mtu value 9000 is used for 10gb interfaces in forums and on the web. but my only connection between the 2 points is with a 10gb interface. all my other connections are 1GB.

Thanks

You don't need to change the MTU to 9000 or anything. Keep the default at 1500. Changing the MTU because you read it in some forum will most likely cause many more problems than it will solve.
Higher than 1500 MTU is for edge cases. For most traffic types 9000 MTU won't give any considerable advantage (considering the disadvantages of troubleshooting the thing).

Also having 9000 MTU only between the switches is pretty much useless. For any effect on this you'd need to use 9000 MTU end to end on all your devices.

Bandwidth test IS a problem if you run it on the switch.
You need to run traffic generator (for capable mikrotik devices) or iperf (for capable linux devices) through the switches. Any other way involving the switches themselves or using the outdated bandwidth test tool will show you bad or simply wrong results.

Chaos has already said this but leave your MTU alone.
Mikrotik bandwidth test is terribly CPU limited and the CPU in those switches is minuscule, chances are you will be running 10G over the link but you are lacking the means to confirm/test it.

thanks to everone...
thanks for sharing yours experience and information with me...

Hi,
Just to share for those who wants to test 10Gbps connection.
I had setup in a remote host our first 10Gbps connection on one of our building this week.
To be able to test it I had to use 10 different sites and push 1Gbps to new location to fill the 10Gbps bandwidth.
Surprisingly I was able to get 10Gbps download but upload would never go over 1Gbps.
After so many tries and discussion with other gigs we come into the light that Mikrotik itself can not generate more than 1Gbps therefor in order to be able to do upload I had to create multiple connection and upload 10 separate connection in order to be able overcome the Bandwidth Test for upload.

The reason I was able to achieve 10Gbps download was because the Test were being generated from different sites and 10Gbps router needed just to respond and show the process.

I hope this help others who wants to do the testing and don't think that provider has an issue. As this was our first impression.

Thanks
Rapidcom