MikroTik CCR2004 Performance Test Part 1

[Music] hi there today we are going to take a look at ccr 2004 so this is the new generation of microtech routers the new generation because it's based on the number two instead of one uh and the differences in processor so mikrotik has a problem because the letter processors that were used in the previous routers are not supported by linux core anymore and that's probably the reason that they chose um another type of processor for the next version of their cloud core routers and that's the arm 64 bit chip appears that it's the one over here with four cores and quite high frequency for me critique case uh at 1.7 gigahertz so this router is running router rise number six we expect it to be running seven soon and it has a clear upgrade pad for the future and we cannot say the same for the tiler processors so the the main point of this presentation is to measure the tripof of this router uh mikrotis is stating here on the website somewhere that it should do somewhere around 26 megabits per second but they do not say how this was tested they just mentioned the equipment they were using for those tests so we're gonna imagine that they had in mind some kind of real performance tests using for example two servers two routers interconnection between the routers and because we don't have the 25 gigabit cable we're gonna instead use two 10 gigabit spf direct access cables and connect them using lacp protocol so that should be a little bit less than 25 gigs but still quite okay so let's take a look at our setup so we have two servers here uh and they are now connected by this spf cables uh from one to the other here you see four ports for two cables so we'll do the first speed test like this and then we'll switch so that the first server will be connected to the first router and we already have this interconnection between those two routers in place and then the second server will be connected to the second router like into these ports here so let's test the speed now between the servers so that we'll see what's the potential of the biggest speed we can achieve using this connection without the routers first so we are back at the computer and we are now connected via ssh uh to the both servers so first and the second one so it doesn't really matter which is which and we're gonna start iperf 3 as a server on the first one and we're going to start iperf 3 as a client using the other server's address as the second one hopefully this one is the right one and we're gonna use two streams because we are using lacp and one stream would only yield 10 gigs okay we can even try this so this is just with one stream you see that it's running at 9.4 and something gigabits per second okay and now the second time we're gonna use two streams and you see it's going in in two blocks and it's displaying the sum and the sum is actually almost double as before at 18.8 gigabits per second okay that's it that completes the first test and we're gonna change the the cables and run the second test now okay let's connect the cables so we should disconnect the cable somewhere here let's take this one and this one and we'll connect them here luckily it doesn't matter which got one go switch we just have to turn them in the right side okay great now we need another pair of cables yeah they are here and that's probably like this no like that cool connect it if this leads up then it's okay this and this one goes like this yeah and like that over here super okay so this is all connected so i already configured those two they are in lacp modes so this is the first rank this is the second trunk and again the first rank on the second router and the second trunk on this second router and they are all in bridge mode now without using any of fastpad technology or something like this just normal bridge so let's see how it's all set up on the computer this is the first server it's called pve and the ports are named no7 and no 8 and they are bound together using this bond zero with lacp protocol and we are using hashing policy layer three and four why we're using this because we will run speed test using hyperf and of course the the mac addresses and ip addresses of those servers will be always be the same so that's why we need to look at the layer 4 so that's basically a source port of the stream that's gonna be made there and if we wouldn't do it like this then uh lacp would only use one channel because uh the hashing algorithm wouldn't distribute the the traffic over to uh parallel lines and it would just use one of course so this is then bound to the to the second bridge the first bridge is just management and we haven't signed it an ip address so we can do the the the ip on tcp ip traffic over here uh nothing special really here and the second server is configured the same way so the bond is using the same ports on the other server of course the same lacp protocol and this bridge is also configured the same way using another ip in the same block okay so if we now start um if we now try and see how how this goes let's go to this server so i'll just take a look of what address is it so this one is number 201 so i can ping the other one 202 so and it's pinging so we can start uh iperf server on on the other one and run the iperf client from this one so i will use hyper version 3 because it introduces the multiple chain channels so it takes like if you would run several ip iperf clients at the same time so we don't have to do it manually we're just using this so this is now started in a server mode and the client mode and two channels so you see this is not very impressive so you see what's happening now we are getting two channels one using five and something gigabytes the second one using two and something gigabytes together they're using eight gigabytes whereas we should see something like 18 or 19 gigabytes like we will see when we connect the cables directly from one server to the other okay let's take a look how the micro deck is configured and if we can do some improvements over there so we are now at the windows computer uh the reason is that winbox is only running on windows computers you can also start it on on the macintosh but there are some problems and we don't want to get involved in those problems so um how what we did here so what we're looking at right now is the bonding interfaces so up down is the connection between upper and and lower my critique and this one goes to the server so let's take a look at how it's configured so i think that the problem with speed here is that micro tick is not treating very good the layer 3 and 4 traffic here so maybe if we lower this to just layer two since the servers are already hashing and kind of routing the traffic to those two lines i don't think we have to do that again here so we'll try just to change this to to much simpler layer to hashing algorithms so this is uh now done in at the first router and we'll do the same at the second one so here there too so you see that the lacp port starts right after it's reconfigured okay so this is now done and we can go back to the traffic generators on our supermicroservers so if this is all right we should still be running the iperf server on the other side so we could just repeat this command here so you see the traffic is almost identical so we didn't do much here so what's the conclusion mikrotik can only route 8.3 gigabytes of track traffic simultaneously in both directions using this type of configurations why did they choose those kind of configuration because the real case scenario would probably be to interconnect those routers using the 25 gigabit cable so because we don't have it we emulated this with the lacp which would probably be the the chosen protocol for network administrator in such a such a case where you have a router that's supposed to do 20 plus gigabyte routing and you don't have the 25 giga cable anyway even if we did have the 25 gigabit cable if we take a look at the block diagram of the microtech router we can see that those two 25 gigabit ports are connected to some kind of multiplexer they are calling port extender and this port extender joins 12 10 gigabit sports which are supposedly in a separate switch together with this board this does not say how fat this line is so we are guessing that even in in pure switch mode if we would just switch from 25 gigabit port to let's say two of the 10 gigabit ports we probably would not achieve the whole 25 gigabit speed anyway because of this multiplexer here okay so this concludes our video if the responses are great i'll try to do the routing scenario as well i expect the results there to be quite the same but you never know we'll see thanks for watching and bye

Channel: Virtua Unboxing

Views: 10,979

Keywords:

Id: EdO-ybgZ2bU

Length: 13min 43sec (823 seconds)

Published: Wed Dec 09 2020