When traffic is flowing from vlan13 ( host at port sfp-sfpplus3 ) towards the network on sfp-sfpplus1 ( not bridged, no vlan ) it’s not fasttracked
Although the connection is marked as fasttracked. But the bytecounter of the fasttrack dummy rule is not increasing.
The CPU is hitting 100% ( one core ) at ~500mbit/s
Traffic the other way around is fasttracked as expected. 1Gbit/s linespeed and CPU Is far away from 100%
It seems the problem is traffic coming in on a vlan interface on top of a bridge.
I know that it’s main purpose is switching. But with fasttrack the cpu has no problem to route 1Gbps. So why using a separate device for that? And it’s an CRS not a CCS
Yes i’m routing here. The routeros receives the traffic on a VLAN interface that is on top of the bridge. I’m talking about the routed traffic not the bridged traffic. The bridged ( switched) traffic works fine, it’s hardware offloaded as it should be.
i think mikrotik will end killing CRS line because of situations like this
I think The fact CRS switch have routeros dont imply you have to do routing on it, i think a switch is a switch, and must be used like that, the advantages of having routeros on it comes from management perspective, you have a very powerfull and versatile winbox graphical user interface, integrated graphs…
it’s called cloud router switch. That does not imply that it’s a switch only either. I’m fully aware that it has not the CPU power for line speed routing. But for the 3xx series they have a CPU that has more power than some of the “routers”. And why not utilize those CPU cycles? whats the point of a 2core cpu if you use only the hardware offloaded switching function and the cpu idles at 1%?
Those switches fit perfect where you have a need for a fast local network and you have a somewhat slower internet connection. Putting an OLD RB2011 next to a CRS317 just to route 500Mbps internet connection ( with fasttrack ) is somehow stupid as the CPU much slower.
that’s my opinion. What i would like to know is whether the limitation i run into has a technical reason or they just forgot of the scenario of a vlan on top of the (hardware offloaded) bridge and that they need to add a few lines of code to get such traffic handled by the fast-tracked path.
Can you post full config, there might be a misconfigured rule creating unexpected symptoms that we might not think about, but seeing the config might ring some bells