I’ve just replaced a dying CCR1036 with one of these and the performance is absolutely woeful, to the point I think its faulty. But thought i’d ask here if there are any known issues with this device because it doesn’t make any sense to me
It’s role in this location is to route approx 1.4gbit/s of traffic as a distribution router, with a handful of PPPoE connections. Even a CCR1009 was capable of doing this but was started to get loaded and would spike up to approx 70% CPU usage. Replaced it with a 1036 which has plenty of CPU power and never saw above 20% but it has a fault that causes periodic reboot
So bought a new CCR2004 and this thing is crap, its struggling to push more than 700mbit/s through its main bonded uplink connection but only on TCP. UDP works just fine and will easily saturate all links
CCR2004 is also around the 15-25% CPU mark. Enabling fast track drops it to ~9% but makes no difference
BTest to locally connected devices with enough CPU (such as Cube60 pro) cap out around 800mbit/s on TCP, they start out at 200mbit/s and ramp up over a few seconds. Again 1009/1036 would effortlessly saturate the link and sustain 920mbit/s solidly. UDP has no problem
Is this actually expected behavior? Because this is pathetic and if I swapped it for a RB4011 i’d have better performance and it would effortlessly handle this traffic load, which makes no sense with its substantially slower CPU
Currently running 7.17.2 and will try rolling it down to 7.15.3 after hours incase its a software bug
I’m hoping this isn’t a repeat of the RB3011 series, where for whatever reason it just cannot saturate it’s CPU and will bottleneck TCP traffic in unexpected ways, only solution was to rip out and replace with 4011 or CCR1xxx series