LAN to LAN Performance
-
I have a setup where I have placed my "wired" devices on the LAN1 interface and my "wireless" on the LAN2 interface.
I noticed some issues when using SMB from my server to the wireless clients that i did not see on the wired.
I moved a wired machine to the LAN2 to test.
The rPerf baseline was all on LAN1 between server and client got >900 Mbps as expected. When I moved my client to LAN2 it was only 25 Mbps.
Any idea why this would be?
My Specs:
-
@brephil Have you baselined LAN to WAN yet? Have you tried iperf to pfSense, there is an iperf package. If I'm interpreting that correctly, it looks like hyperthreading is enabled, last I checked it is recommended to turn that off.
What does load on pfSense look like during these tests? CPU time, interrupt rate, anything else that stands out?
My quad-core i5 is getting line-rate full-duplex gigabit WAN-to-LAN with NAT and traffic shaping, while staying under 20% load. Or at least last I cared to check back around version 2.3.
-
I should add... LAN2(or LAN1 at that matter) to WAN is good... Speed Test sees routinely 300-400 Mbps (via wifi) or 600-800 Mbps (when I "plug in" the device).
So really odd see this slow down between to LAN interfaces... LAN1 to/from LAN2 It's so consistent almost like there is a limiter in there, but did not set one up.
If anyone has this general setup and see nominal speeds let me know like to compare notes.
-
@brephil is LAN2 over a different type of network card? Like LAN1 being Intel and LAN2 being realtek or something?
-
@harvy66 LAN1 and LAN2 are ports on the same nic. Intel 4 port card.
The CPU utilization is 10 to 12 %. Not heard the HT recommendation before I will give this a try.
If there a way to run rPerf between WAN and LAN? I have the package installed.
-
@brephil Not sure why this happened, but when removed some interfaces from a Ubuntu server it got better (as in line speed). I think the server was somehow trying to do some routing on it's own as it had access to both LANs.
-
@brephil Maybe it created a loop?
The best test for iperf is not against pfSense, but through it. Put a device in the LAN and another in the WAN and do iperf between those two devices.