Separate or dual NIC card?

[dwasifar]

When building an OPNsense box on standard PC hardware, is there a performance difference between having two separate NICs vs a single dual-port NIC?

[monstermania]

Hi,
it depends on many factors (i.e. the used hw, mainboard, chipsets, etc.) btw. the hw design. Principally all devices into a pc connected by serial pci-e lanes. An pci-e single line offers a speed of 4Gbit (gross). Normaly enough for 2 Gbit lan cards. But if the hw vendor connects some more devices to the same pci-e line all devices had to share the 4 Gbit...
Same thing for lan card nics. Good lan chipsets have their own data processors build in. So they don't use much host pc cpu power.
Cheap chipsets (i.e. Realtek) needs more host cpu power for handle the same network load.

So your simple question isn't easy to answer!
If you need specific lan speed the simplest solution is to buy an preconfigured appliance where the manufacturer quarantees you the speed.

best regards
Dirk

[dwasifar]

Thanks for the detailed reply.

I don't have a specific performance requirement number; I only want to get the best performance out of the hardware I have. 

I didn't know that about processors on the NIC, but it makes sense.

The question is mainly curiosity.  My actual firewall appliance is a Watchguard 5 series that's currently running pfSense.  I'm switching it to OPNsense, so to avoid downtime I built a temporary firewall appliance running OPNsense on an old PC with two Intel NICs.  I expected to see a performance hit from having two separate NICs, but that didn't turn out to be the case; throughput on the temporary appliance is the same as it was on the Watchguard.

But, as you say, there are a lot of variables here.  There are no other PCI-E devices plugged in, but I don't know if on-motherboard devices share the line. 

I suppose the right way to go about finding out would be to test a dual NIC against a pair of single NICs, each with the same chipset, in the same system, and do benchmarks.