I am trying to understand or see if there would be differences between using OpenBSD for routing in a setup where the routing is done between two VLan's for example oppose to between to physical network cards.

Any impact on the pps capability between the two?

Internally to the server/router, is the processor actually look at all the packets content, or just the headers and the rest is done via DMA, or what else?

I would to find more informations as to what part play what in routing in the OpenBSD system.

I understand that it is more the pps that is a bottleneck in routing, so this makes me think that only the headers are then inspected. So, if physical limitations inside hardware like PCI bus for example, then may be routing between VLan's on the same network card might have different impact. But I may not think about this the right way here. Would the process of routing be the same cost on the CPU and suffer the same limitations here?

In theory if only the header for the VLan would need to be change and somehow the network could then forward the rest of the packet as is with the new header without the need to cross over some internal server PCI bridge for example, then the performance should definitely be much better, but am I thinking about this the wrong way, or am I so stupid that this doesn't make any sense?

I would love if someone could put more lights on this for me and explain it in ways that I could fully understand the internal process done in the server for that. What parts are done by what and witch one are suffering of limitations, when done inside OpenBSD.

If I could spare some of your time, I would love to understand it, or if you can refer to some possible good explications on the net for that, I would very much appreciate it!

Thanks a millions.

Daniel

Reply via email to