I just purchased CCR2004-PC as replacement to my previous CCR1009-PC device. In CCR1009-PC and CCR1009-RM there was 200 mhz difference between rackmount and desktop version (1000 vs 1200 mhz). With CCR2004 there’s 300 mhz difference. In CCR1009-PC I could change CPU frequency back to 1200 mhz using
/system health set cpu-frequency=1200mhz
How can I achieve similar result with CCR2004?
You cannot. The PC version is locked at 1200 MHz.
So it’s device that realistically achieves lower performance than CCR1009-PC at 1200 mhz which it’s supposed to replace?
And as a bonus we get no LCD screen, no beeper, no usb, no sd card slot and no smart card reader?
cool. Awesome even.
As always one should do his/her homework before purchasing. All the details are provided upfront by Mikrotik. The non-+PC CCR2004 has a 1700 MHz CPU. The +PC CPU had to be dialed back a little due to being passively cooled.
All that said, when you disider the CCR2004 is a 4 core device instead of 9, it performs quite nicely. My recollection is that the CCR2004 excels at single core performance. Also, TILE is dead so that’s a sinking platform anyway. Mikrotik has discontinued all their TILE devices. Can’t speak to all the other features you mention. I personally have no use for any of them. Again though, this is all documented up front so it’s not like Mikrotik deceived you.
always is important to check datasheet, specs etc, preferably,
before purchase
you will avoid a lot of regrets
comparison of both passive devices performance looks very simmilar, not an upgrade in some scenarios but enough if you need a replacement
Yeah but spec sheet does not provide information whether device can be overclocked or not. All older MikroTiks did allow rising CPU clocks. It’s not specified in spec sheet that CCR1009 passive is able to run at 1200 mhz as well yet it can. And then it has higher performance than CCR2004 passive (due to its inability to match rackmount clocks). Basing on my past experience I had no reason to assume it’s not possible to adjust clock speed of CCR2004 to like 1400 at least.
My story is that I thought my CCR1009-PC died and ordered replacement CCR2004-PC unit. After few days of troubleshooting I managed to bring that CCR1009 back to life but CCR2004 already arrived. I decided that well… since I already have more recent device I may just as well just migrate to it. It will probably last longer…
But it turns out that CCR2004 is actually inferior device comparing to CCR1009-PC (considering that I’m running it at 1200 mhz) so it’s kinda ridiculous situation where it doesn’t make logical sense to migrate to new device… I hope I’ll be able to return that CCR2004…
So it’s your fault for expecting to be able to overclock a device. If you’re overclocking often, then you should know by now that some devices can be overclocked nicely but others can’t be overclocked at all … and without trying to do it (either yourself or by some other person which then shares experiences) you don’t know how it’ll go.
Mhz is not a measure of performance. The ARM64 in CCR2004 is far more powerful than the old TILEs.
You buy the PC version because you need either passive cooling or DC-input. If you need performance should choose the AC version.
The LCD display was just a waste of money and SD is not a serious storage medium.
Well I wouldn’t bring overclocking argument to the table at all if CCR2004-PC provided higher base performance out of the box than CCR1009-PC at 1200mhz in the first place. Basically my point is that with fairly basic config CCR1009-PC is faster (marginally but still) device than CCR2004-PC that is supposed to be its successor. So it’d be rational to expect it to be just plain faster. Period. It’s just not sane thing to assume that successor of device model will perform worse. It didn’t even come across my mind that such situation could exist in real world.
That said I checked pricing of CCR1009-7G-1C-1S+PC back in the day, corrected it by inflation and compared to CCR2004, coming to conclusion that in matter of fact CCR2004-PC is realistically almost 50% cheaper than CCR1009-PC was at the time (considering value of money) so… idk. I guess I can’t complain as much as I thought I can…
I know, I was comparing benchmark results not clock-to-clock. I was comparing CCR1009-RM benchmark results - which has 1200 mhz ootb because you can set CCR1009-PC to 1200mhz as well. That’s why I used that table for comparison to CCR2004-PC. So I basically compared best case CCR1009-PC to best case CCR2004-PC. And in such case CCR1009-PC is (at least according to benchmarks) faster device than CCR2004-PC.
But honestly I guess I’m in general just salty because I’m very disappointed with how… well, disappointing CCR2004-PC is as device. Even if you compare it to like RB5009UG+S+IN that is over 50% cheaper. CCR2004-PC is very anemic yet expensive device in lineup (the same price as RM variant that has significantly higher performance AND dual PSU). If I remember correctly CCR1009-PC was also priced the same as CCR1009-RM but then it made sense - since you could set the same clock so it could essentially achieve the same performance and it was kind of the same hardware. Now it’s not. It’s significantly inferior to RM variant for the same price. So there’s plenty of reasons to be upset ngl, no matter which way you look at it - it’s underwhelming.
Underwhelming for value comparing to RB5009/RB4011,
underwhelming comparing to its RM counterpart,
underwhelming comparing to its predecessor.
Well yes. Maybe indeed I just got a little bit misguided by claim of “Up to 300% faster than the previous CCR1009 routers” without adding “And up to 57% slower than the previous CCR1009 routers”
Hmm. Now when I think about it - not all features are documented upfront. This is block diagram of CRS317:
https://i.mt.lv/cdn/product_files/CRS317-1G-16S_220904.png
It does not mention that this device does in fact have beeper (that I use). I have 3 of those switches and all of them feature beepers that I use for signaling issues with connectivity. So it’s not always trivial to find out whether certain device has certain features.
Also like… CCR1009 is discontinued anyways and there’s nothing else comparable really (with any features that I’m missing). So atm really all I can do is complain, cry in the corner and be sad about current situation. If I would be very stubborn I could probably buy RM variant and swap boards but that sounds like quite extreme dedication…
The spec section tells you it does.
oh… okay. It used to be put on block diagrams in the past. That’s why I looked there.
https://i.mt.lv/cdn/product_files/CCR1009-7G-1C-1SplusPC_200618.png
But fair point.
I personally really enjoy my CCR2004 +PC router. It works wonderfully for my needs. I’d be really curious to see the difference in performance between devices for you if any. CPU/RAM usage, throughput, et. Data driven debate is my favorite since data doesn’t lie and doesn’t care about feelings.
Also I noticed that the CCR2004 has significantly more ports including an additional SFP+, double the RAM, more modern and efficient processor type - we’re not really comparing oranges to oranges in this instance.
RB5009UG is a very different device, it’s half the ports.
We find the CCR2004-PC very useful and performant. No issues at all.
interesting topic on how to blame others about your bad purchase decisions 
There are a lot of them on this forum. Bad hardware, bad performance, bugs… And then it turns out is a misconfiguration, bad cable, or device that wasn’t appropriate for the specific needs.
You may not remember that but you specifically told me to NOT buy CCR2004-PC. I just found this thread which I completely forgot about XD
Thread can be found here:
http://forum.mikrotik.com/t/ccr2004-1g-12s-2xs-are-there-any-before-you-buy-caveats/163377/1
EDIT: srr, it was @sirbryan who told to NOT get passively cooled one xD
As a result I have CCR1009-PC, CCR2004-PC and CCR2004-12S+ and instead of those two CCR2004’s I could probably just get CCR2116 for price of those and get better results XD (of course I only need one router)
I’m not entirely blaming for bad purchase as choice because there’s nothing else really that I could/would buy instead (since I kinda need this form factor anyways). I’m just salty that CCR2004-PC doesn’t have higher performance. (I don’t think I would buy RB5009UG which seems to be the only alternative because I’m using two optical uplinks)
Since it’d be cool if this thread ended with some… more productive conclusion than hurr durr CCR2004-PC sucks I’ll try to perform some benchmarks for my use case before I return CCR2004-PC and post results here. CCR1009-PC vs CCR2004-PC vs CCR2004-12S+ (which should perform more or less the same as CCR2004-RM for my use case).
I really look forward to seeing the test results based on your use case. Perhaps the CCR2004+PC performance will surprise you considering the differences that have been noted and discussed above. The fact that you’ve taken your lumps from some of us and want to improve the tone of this thread is admirable and refreshing. Learning, even if through error, is fun!
Okay so I’m back.
Yes it did. And in fact for my use case difference between 12S+ variant fall as low as 8% which while still makes me a little bit salty, is still considerable improvement from claimed 15% drop indicated by benchmarks on website. That was TL;DR version, now into details:
I don’t think I’ll give more than minimal attention to CCR1009 performance in this comparison because it got so obliterated by both CCR2004’s that I didn’t even really feel like testing it beyon initial run.
root@raid /h/lapsio [1]# iperf3 -c 192.168.8.81 -p 2000 -P 1 -t 300 -l 1M
Connecting to host 192.168.8.81, port 2000
[ 5] local 192.168.8.82 port 37920 connected to 192.168.8.81 port 2000
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 457 MBytes 3.83 Gbits/sec 818 489 KBytes
[ 5] 1.00-2.00 sec 441 MBytes 3.70 Gbits/sec 761 1014 KBytes
[ 5] 2.00-3.00 sec 460 MBytes 3.86 Gbits/sec 872 699 KBytes
[ 5] 3.00-4.00 sec 452 MBytes 3.80 Gbits/sec 987 926 KBytes
[ 5] 4.00-5.00 sec 451 MBytes 3.78 Gbits/sec 1101 909 KBytes
[ 5] 5.00-6.00 sec 457 MBytes 3.83 Gbits/sec 1459 760 KBytes
[ 5] 6.00-7.00 sec 457 MBytes 3.83 Gbits/sec 1366 865 KBytes
[ 5] 7.00-8.00 sec 431 MBytes 3.62 Gbits/sec 545 1.01 MBytes
[ 5] 8.00-9.00 sec 469 MBytes 3.93 Gbits/sec 853 760 KBytes
[ 5] 9.00-10.00 sec 429 MBytes 3.60 Gbits/sec 845 900 KBytes
[ 5] 10.00-11.00 sec 456 MBytes 3.82 Gbits/sec 1091 743 KBytes
[ 5] 11.00-12.00 sec 412 MBytes 3.46 Gbits/sec 654 690 KBytes
[ 5] 12.00-13.00 sec 474 MBytes 3.98 Gbits/sec 1206 594 KBytes
[ 5] 13.00-14.00 sec 474 MBytes 3.97 Gbits/sec 1382 839 KBytes
[ 5] 14.00-15.00 sec 453 MBytes 3.80 Gbits/sec 957 577 KBytes
[ 5] 15.00-16.00 sec 453 MBytes 3.80 Gbits/sec 989 979 KBytes
^C[ 5] 16.00-16.60 sec 292 MBytes 4.08 Gbits/sec 882 865 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-16.60 sec 7.34 GBytes 3.80 Gbits/sec 16768 sender
[ 5] 0.00-16.60 sec 0.00 Bytes 0.00 bits/sec receiver
iperf3: interrupt - the client has terminated
Those are results well known to me, result that I have seen long before and results that were far from satisfying for me since my machines were always connected by 10G link. Yeah sure, CCR1009 was in fact able to hit 10G but it required using fastpath and that simply wasn’t an option in my setup. It’s 9000 jumbo result. With normal frames it barely exceeded 1G. It’s 100% single core use case and router was obviously hammered on single core to 100%. After that initial benchmark I copied core parts of config to CCR2004-12S+ and re-run the same tests with following result:
root@raid /h/lapsio [1]# iperf3 -c 192.168.8.81 -p 2000 -P 1 -t 300 -l 1M
Connecting to host 192.168.8.81, port 2000
[ 5] local 192.168.8.82 port 47546 connected to 192.168.8.81 port 2000
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 1.11 GBytes 9.54 Gbits/sec 70 2.24 MBytes
[ 5] 1.00-2.00 sec 1.12 GBytes 9.61 Gbits/sec 10 2.53 MBytes
[ 5] 2.00-3.00 sec 1.11 GBytes 9.55 Gbits/sec 13 2.76 MBytes
[ 5] 3.00-4.00 sec 1.12 GBytes 9.60 Gbits/sec 10 2.06 MBytes
[ 5] 4.00-5.00 sec 1.13 GBytes 9.72 Gbits/sec 7 2.20 MBytes
[ 5] 5.00-6.00 sec 1.12 GBytes 9.61 Gbits/sec 2 2.48 MBytes
[ 5] 6.00-7.00 sec 1.13 GBytes 9.74 Gbits/sec 9 2.61 MBytes
[ 5] 7.00-8.00 sec 1.13 GBytes 9.71 Gbits/sec 11 2.73 MBytes
[ 5] 8.00-9.00 sec 1.12 GBytes 9.63 Gbits/sec 7 2.04 MBytes
[ 5] 9.00-10.00 sec 1.13 GBytes 9.70 Gbits/sec 14 2.19 MBytes
[ 5] 10.00-11.00 sec 1.12 GBytes 9.62 Gbits/sec 6 2.55 MBytes
[ 5] 11.00-12.00 sec 1.13 GBytes 9.68 Gbits/sec 5 2.76 MBytes
[ 5] 12.00-13.00 sec 1.12 GBytes 9.60 Gbits/sec 12 2.12 MBytes
[ 5] 13.00-14.00 sec 1.13 GBytes 9.70 Gbits/sec 6 2.28 MBytes
[ 5] 14.00-15.00 sec 1.12 GBytes 9.66 Gbits/sec 4 2.50 MBytes
[ 5] 15.00-16.00 sec 1.13 GBytes 9.69 Gbits/sec 6 2.69 MBytes
[ 5] 16.00-17.00 sec 1.12 GBytes 9.63 Gbits/sec 4 2.93 MBytes
[ 5] 17.00-18.00 sec 1.13 GBytes 9.70 Gbits/sec 9 2.24 MBytes
[ 5] 18.00-19.00 sec 1.11 GBytes 9.56 Gbits/sec 18 2.78 MBytes
[ 5] 19.00-20.00 sec 1.12 GBytes 9.58 Gbits/sec 89 2.76 MBytes
[ 5] 20.00-21.00 sec 1.13 GBytes 9.74 Gbits/sec 13 2.69 MBytes
[ 5] 21.00-22.00 sec 1.13 GBytes 9.69 Gbits/sec 6 2.85 MBytes
[ 5] 22.00-23.00 sec 1.14 GBytes 9.80 Gbits/sec 7 2.82 MBytes
^C[ 5] 23.00-23.63 sec 735 MBytes 9.80 Gbits/sec 3 2.74 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-23.63 sec 26.6 GBytes 9.67 Gbits/sec 341 sender
[ 5] 0.00-23.63 sec 0.00 Bytes 0.00 bits/sec receiver
iperf3: interrupt - the client has terminated
Which was quite okay but felt like device is on the edge and indeed CPU usage looked like this:
[lapsio@CCR2004SWAG] /ip/firewall/nat> /system/resource/cpu/print
Columns: CPU, LOAD, IRQ, DISK
# CPU LOAD IRQ DISK
0 cpu0 0% 0% 0%
1 cpu1 0% 0% 0%
2 cpu2 100% 100% 0%
3 cpu3 0% 0% 0%
So being kinda worried that CCR2004-PC would fall below 10G if CCR2004-12S+ barely reaches it, I proceeded to test CCR2004-PC
root@raid /h/lapsio# iperf3 -c 192.168.8.81 -p 2000 -P 1 -t 300 -l 1M
Connecting to host 192.168.8.81, port 2000
[ 5] local 192.168.8.82 port 54428 connected to 192.168.8.81 port 2000
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 1.14 GBytes 9.80 Gbits/sec 34 1.91 MBytes
[ 5] 1.00-2.00 sec 1.15 GBytes 9.86 Gbits/sec 21 1.67 MBytes
[ 5] 2.00-3.00 sec 1.15 GBytes 9.86 Gbits/sec 16 1.70 MBytes
[ 5] 3.00-4.00 sec 1.15 GBytes 9.86 Gbits/sec 20 1.55 MBytes
[ 5] 4.00-5.00 sec 1.15 GBytes 9.84 Gbits/sec 22 1.19 MBytes
[ 5] 5.00-6.00 sec 1.15 GBytes 9.85 Gbits/sec 16 1.25 MBytes
[ 5] 6.00-7.00 sec 1.15 GBytes 9.87 Gbits/sec 23 987 KBytes
[ 5] 7.00-8.00 sec 1.15 GBytes 9.86 Gbits/sec 17 1.18 MBytes
[ 5] 8.00-9.00 sec 1.15 GBytes 9.86 Gbits/sec 22 1.73 MBytes
[ 5] 9.00-10.00 sec 1.15 GBytes 9.86 Gbits/sec 33 1.80 MBytes
[ 5] 10.00-11.00 sec 1.15 GBytes 9.87 Gbits/sec 15 1.62 MBytes
[ 5] 11.00-12.00 sec 1.15 GBytes 9.86 Gbits/sec 23 1.85 MBytes
[ 5] 12.00-13.00 sec 1.15 GBytes 9.87 Gbits/sec 21 1.75 MBytes
[ 5] 13.00-14.00 sec 1.15 GBytes 9.86 Gbits/sec 21 1.37 MBytes
[ 5] 14.00-15.00 sec 1.15 GBytes 9.86 Gbits/sec 19 1.74 MBytes
[ 5] 15.00-16.00 sec 1.15 GBytes 9.87 Gbits/sec 18 1.71 MBytes
[ 5] 16.00-17.00 sec 1.15 GBytes 9.87 Gbits/sec 21 1.72 MBytes
[ 5] 17.00-18.00 sec 1.15 GBytes 9.87 Gbits/sec 17 1.20 MBytes
[ 5] 18.00-19.00 sec 1.15 GBytes 9.86 Gbits/sec 48 1.71 MBytes
[ 5] 19.00-20.00 sec 1.15 GBytes 9.88 Gbits/sec 28 1.20 MBytes
[ 5] 20.00-21.00 sec 1.15 GBytes 9.87 Gbits/sec 26 1.22 MBytes
[ 5] 21.00-22.00 sec 1.15 GBytes 9.87 Gbits/sec 20 1.17 MBytes
^C[ 5] 22.00-22.23 sec 267 MBytes 9.78 Gbits/sec 9 1.23 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-22.23 sec 25.5 GBytes 9.87 Gbits/sec 510 sender
[ 5] 0.00-22.23 sec 0.00 Bytes 0.00 bits/sec receiver
iperf3: interrupt - the client has terminated
And I don’t f*cking get it. CPU load was also 99-100% but transfers were much more stable and higher despite lower CPU clocks. But that could be just result of PIPE thingy on 12S+ so I decided to go lower with MTU, to 1500:
CCR2004-12S+
root@raid /h/lapsio [1]# iperf3 -c 192.168.8.81 -p 2000 -P 1 -t 300 -l 1M
Connecting to host 192.168.8.81, port 2000
[ 5] local 192.168.8.82 port 36282 connected to 192.168.8.81 port 2000
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 242 MBytes 2.03 Gbits/sec 104 419 KBytes
[ 5] 1.00-2.00 sec 242 MBytes 2.03 Gbits/sec 24 399 KBytes
[ 5] 2.00-3.00 sec 241 MBytes 2.02 Gbits/sec 53 462 KBytes
[ 5] 3.00-4.00 sec 242 MBytes 2.03 Gbits/sec 95 468 KBytes
[ 5] 4.00-5.00 sec 242 MBytes 2.03 Gbits/sec 79 414 KBytes
[ 5] 5.00-6.00 sec 241 MBytes 2.02 Gbits/sec 49 447 KBytes
[ 5] 6.00-7.00 sec 242 MBytes 2.03 Gbits/sec 54 365 KBytes
[ 5] 7.00-8.00 sec 241 MBytes 2.02 Gbits/sec 14 441 KBytes
[ 5] 8.00-9.00 sec 240 MBytes 2.01 Gbits/sec 25 495 KBytes
[ 5] 9.00-10.00 sec 243 MBytes 2.04 Gbits/sec 63 486 KBytes
[ 5] 10.00-11.00 sec 241 MBytes 2.02 Gbits/sec 11 385 KBytes
[ 5] 11.00-12.00 sec 242 MBytes 2.03 Gbits/sec 82 320 KBytes
[ 5] 12.00-13.00 sec 242 MBytes 2.03 Gbits/sec 63 438 KBytes
[ 5] 13.00-14.00 sec 240 MBytes 2.01 Gbits/sec 35 431 KBytes
[ 5] 14.00-15.00 sec 242 MBytes 2.03 Gbits/sec 7 438 KBytes
[ 5] 15.00-16.00 sec 241 MBytes 2.02 Gbits/sec 33 421 KBytes
[ 5] 16.00-17.00 sec 240 MBytes 2.01 Gbits/sec 65 431 KBytes
[ 5] 17.00-18.00 sec 241 MBytes 2.02 Gbits/sec 75 321 KBytes
[ 5] 18.00-19.00 sec 241 MBytes 2.02 Gbits/sec 64 337 KBytes
[ 5] 19.00-20.00 sec 240 MBytes 2.01 Gbits/sec 9 436 KBytes
[ 5] 20.00-21.00 sec 244 MBytes 2.05 Gbits/sec 20 450 KBytes
^C[ 5] 21.00-21.46 sec 112 MBytes 2.04 Gbits/sec 24 338 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-21.46 sec 5.06 GBytes 2.03 Gbits/sec 1048 sender
[ 5] 0.00-21.46 sec 0.00 Bytes 0.00 bits/sec receiver
iperf3: interrupt - the client has terminated
CCR2004-PC
root@raid /h/lapsio [1]# iperf3 -c 192.168.8.81 -p 2000 -P 1 -t 300 -l 1M
Connecting to host 192.168.8.81, port 2000
[ 5] local 192.168.8.82 port 40322 connected to 192.168.8.81 port 2000
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 354 MBytes 2.97 Gbits/sec 47 440 KBytes
[ 5] 1.00-2.00 sec 348 MBytes 2.92 Gbits/sec 20 380 KBytes
[ 5] 2.00-3.00 sec 350 MBytes 2.94 Gbits/sec 50 457 KBytes
[ 5] 3.00-4.00 sec 352 MBytes 2.95 Gbits/sec 25 451 KBytes
[ 5] 4.00-5.00 sec 351 MBytes 2.94 Gbits/sec 18 444 KBytes
[ 5] 5.00-6.00 sec 352 MBytes 2.95 Gbits/sec 36 443 KBytes
[ 5] 6.00-7.00 sec 354 MBytes 2.97 Gbits/sec 13 379 KBytes
[ 5] 7.00-8.00 sec 350 MBytes 2.94 Gbits/sec 19 450 KBytes
[ 5] 8.00-9.00 sec 353 MBytes 2.96 Gbits/sec 49 324 KBytes
[ 5] 9.00-10.00 sec 358 MBytes 3.00 Gbits/sec 58 443 KBytes
[ 5] 10.00-11.00 sec 353 MBytes 2.96 Gbits/sec 29 519 KBytes
[ 5] 11.00-12.00 sec 358 MBytes 3.00 Gbits/sec 23 436 KBytes
[ 5] 12.00-13.00 sec 358 MBytes 3.00 Gbits/sec 28 477 KBytes
[ 5] 13.00-14.00 sec 352 MBytes 2.95 Gbits/sec 18 478 KBytes
[ 5] 14.00-15.00 sec 344 MBytes 2.89 Gbits/sec 60 444 KBytes
[ 5] 15.00-16.00 sec 360 MBytes 3.02 Gbits/sec 25 460 KBytes
[ 5] 16.00-17.00 sec 353 MBytes 2.96 Gbits/sec 27 452 KBytes
[ 5] 17.00-18.00 sec 358 MBytes 3.00 Gbits/sec 36 526 KBytes
[ 5] 18.00-19.00 sec 351 MBytes 2.94 Gbits/sec 29 436 KBytes
^C[ 5] 19.00-19.69 sec 244 MBytes 2.96 Gbits/sec 36 423 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-19.69 sec 6.79 GBytes 2.96 Gbits/sec 646 sender
[ 5] 0.00-19.69 sec 0.00 Bytes 0.00 bits/sec receiver
iperf3: interrupt - the client has terminated
Bruh.
After some digging around I realized that it’s most likely down to difference between RouterOS version. Both devices were with factory ROS version but since CCR2004-PC was much more recent, it had 7.12 while CCR2004-12+ had something like 7.5 or 7.8 I don’t remember. Anyhow - after upgrading both routers performance of CCR2004-12+ significantly improved. Results with 9k jumbo now were almost 1:1 the same as CCR2004-PC and 1500 frames results flipped:
CCR2004-12S+
root@raid /h/lapsio# iperf3 -c 192.168.8.81 -p 2000 -P 1 -t 30 -l 1M
Connecting to host 192.168.8.81, port 2000
[ 5] local 192.168.8.82 port 57904 connected to 192.168.8.81 port 2000
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 303 MBytes 2.54 Gbits/sec 95 396 KBytes
[ 5] 1.00-2.00 sec 300 MBytes 2.52 Gbits/sec 70 416 KBytes
[ 5] 2.00-3.00 sec 302 MBytes 2.53 Gbits/sec 69 402 KBytes
[ 5] 3.00-4.00 sec 298 MBytes 2.50 Gbits/sec 26 403 KBytes
[ 5] 4.00-5.00 sec 301 MBytes 2.52 Gbits/sec 46 402 KBytes
[ 5] 5.00-6.00 sec 300 MBytes 2.52 Gbits/sec 86 403 KBytes
[ 5] 6.00-7.00 sec 300 MBytes 2.52 Gbits/sec 71 409 KBytes
[ 5] 7.00-8.00 sec 300 MBytes 2.52 Gbits/sec 47 393 KBytes
[ 5] 8.00-9.00 sec 299 MBytes 2.51 Gbits/sec 31 441 KBytes
[ 5] 9.00-10.00 sec 300 MBytes 2.52 Gbits/sec 25 436 KBytes
[ 5] 10.00-11.00 sec 300 MBytes 2.52 Gbits/sec 48 447 KBytes
[ 5] 11.00-12.00 sec 301 MBytes 2.52 Gbits/sec 72 448 KBytes
[ 5] 12.00-13.00 sec 300 MBytes 2.52 Gbits/sec 87 407 KBytes
[ 5] 13.00-14.00 sec 299 MBytes 2.51 Gbits/sec 59 392 KBytes
[ 5] 14.00-15.00 sec 301 MBytes 2.52 Gbits/sec 61 389 KBytes
[ 5] 15.00-16.00 sec 300 MBytes 2.52 Gbits/sec 25 328 KBytes
[ 5] 16.00-17.00 sec 299 MBytes 2.51 Gbits/sec 58 334 KBytes
[ 5] 17.00-18.00 sec 300 MBytes 2.52 Gbits/sec 61 366 KBytes
[ 5] 18.00-19.00 sec 300 MBytes 2.52 Gbits/sec 29 417 KBytes
[ 5] 19.00-20.00 sec 300 MBytes 2.52 Gbits/sec 53 441 KBytes
[ 5] 20.00-21.00 sec 299 MBytes 2.51 Gbits/sec 36 440 KBytes
[ 5] 21.00-22.00 sec 299 MBytes 2.51 Gbits/sec 48 406 KBytes
[ 5] 22.00-23.00 sec 298 MBytes 2.50 Gbits/sec 36 397 KBytes
[ 5] 23.00-24.00 sec 299 MBytes 2.51 Gbits/sec 85 436 KBytes
[ 5] 24.00-25.00 sec 300 MBytes 2.52 Gbits/sec 46 392 KBytes
[ 5] 25.00-26.00 sec 299 MBytes 2.51 Gbits/sec 51 396 KBytes
[ 5] 26.00-27.00 sec 299 MBytes 2.51 Gbits/sec 33 433 KBytes
[ 5] 27.00-28.00 sec 298 MBytes 2.50 Gbits/sec 56 477 KBytes
[ 5] 28.00-29.00 sec 300 MBytes 2.52 Gbits/sec 57 328 KBytes
[ 5] 29.00-30.00 sec 299 MBytes 2.51 Gbits/sec 22 426 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-30.00 sec 8.78 GBytes 2.52 Gbits/sec 1589 sender
[ 5] 0.00-30.00 sec 8.78 GBytes 2.51 Gbits/sec receiver
CCR2004-PC
root@raid /h/lapsio# iperf3 -c 192.168.8.81 -p 2000 -P 1 -t 30 -l 1M
Connecting to host 192.168.8.81, port 2000
[ 5] local 192.168.8.82 port 48710 connected to 192.168.8.81 port 2000
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 282 MBytes 2.36 Gbits/sec 17 315 KBytes
[ 5] 1.00-2.00 sec 281 MBytes 2.36 Gbits/sec 48 416 KBytes
[ 5] 2.00-3.00 sec 277 MBytes 2.32 Gbits/sec 20 379 KBytes
[ 5] 3.00-4.00 sec 268 MBytes 2.25 Gbits/sec 51 372 KBytes
[ 5] 4.00-5.00 sec 275 MBytes 2.31 Gbits/sec 72 382 KBytes
[ 5] 5.00-6.00 sec 281 MBytes 2.36 Gbits/sec 49 396 KBytes
[ 5] 6.00-7.00 sec 283 MBytes 2.37 Gbits/sec 39 334 KBytes
[ 5] 7.00-8.00 sec 276 MBytes 2.32 Gbits/sec 92 416 KBytes
[ 5] 8.00-9.00 sec 276 MBytes 2.32 Gbits/sec 28 383 KBytes
[ 5] 9.00-10.00 sec 279 MBytes 2.34 Gbits/sec 27 417 KBytes
[ 5] 10.00-11.00 sec 274 MBytes 2.30 Gbits/sec 44 399 KBytes
[ 5] 11.00-12.00 sec 278 MBytes 2.33 Gbits/sec 28 390 KBytes
[ 5] 12.00-13.00 sec 276 MBytes 2.32 Gbits/sec 37 457 KBytes
[ 5] 13.00-14.00 sec 274 MBytes 2.30 Gbits/sec 60 448 KBytes
[ 5] 14.00-15.00 sec 286 MBytes 2.40 Gbits/sec 42 437 KBytes
[ 5] 15.00-16.00 sec 272 MBytes 2.28 Gbits/sec 63 479 KBytes
[ 5] 16.00-17.00 sec 272 MBytes 2.28 Gbits/sec 44 411 KBytes
[ 5] 17.00-18.00 sec 286 MBytes 2.40 Gbits/sec 87 451 KBytes
[ 5] 18.00-19.00 sec 272 MBytes 2.28 Gbits/sec 72 404 KBytes
[ 5] 19.00-20.00 sec 272 MBytes 2.28 Gbits/sec 84 386 KBytes
[ 5] 20.00-21.00 sec 280 MBytes 2.35 Gbits/sec 20 437 KBytes
[ 5] 21.00-22.00 sec 281 MBytes 2.36 Gbits/sec 36 428 KBytes
[ 5] 22.00-23.00 sec 288 MBytes 2.42 Gbits/sec 35 529 KBytes
[ 5] 23.00-24.00 sec 282 MBytes 2.37 Gbits/sec 51 454 KBytes
[ 5] 24.00-25.00 sec 270 MBytes 2.27 Gbits/sec 68 368 KBytes
[ 5] 25.00-26.00 sec 280 MBytes 2.35 Gbits/sec 35 397 KBytes
[ 5] 26.00-27.00 sec 274 MBytes 2.30 Gbits/sec 25 392 KBytes
[ 5] 27.00-28.00 sec 282 MBytes 2.37 Gbits/sec 46 505 KBytes
[ 5] 28.00-29.00 sec 268 MBytes 2.25 Gbits/sec 68 496 KBytes
[ 5] 29.00-30.00 sec 275 MBytes 2.31 Gbits/sec 62 469 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-30.00 sec 8.12 GBytes 2.33 Gbits/sec 1450 sender
[ 5] 0.00-30.00 sec 8.12 GBytes 2.33 Gbits/sec receiver
So it seems that CCR2004-12S+ was quite quirky device that suffered some significant performance issues. Though I’m kind of sad that CCR2004-PC has lower performance with most recent ROS version and performs noticably worse than with factory image - I believe it may be just related to all-round performance rebalancing and optimizing more common use cases than mine.
Overall I’m observing 2.3 Gbps vs 2.5 Gbps difference between CCR2004-12S+ and CCR2004-PC which is not all that bad. It’d be cool if it was smaller difference but I guess it is what it is. It still obliterates CCR1009. So I guess I won’t be returning it after all. Instead I’m gonna proceed to cry why it’s not 2.5 Gbps in use case that isn’t even my actual use case because I apparently have 9k jumbo everywhere in network so… welp.
I honsetly thought it’s not gonna reach 10G in full software with no fasttrack or anything and I thought I’ll have to go for CCR2116. So it’s quite pleasant surprise. Now I’m gonna go for bunch of Intel E810 and pursue 100G in full software XD
Dude, this was an awesome post!! So much data!!! I had faith the newer architecture would outperform the older tile-based models but it’s hard to really compare without real world testing. As far as the performance between the two CCR2004 models, remember that the +PC is running at 1200 MHz while the other is running at 1700 MHz. So it makes sense that it would edge out the +PC model by a bit. Still pretty great performance by the +PC. ROS is very actively under development still so it’s entirely possible that further optimizations and improvements will be made. There have been a number of tweaks to the switch performance in the devices, as well as throughput and stability. I also wonder if you could eek out additional performance with tweaks to your configuration - assuming you might have some inefficiencies (like redundant firewall rules etc). But all in all, it looks like you’ve been won over even if you are a little bit salty in some aspects?