Sudden weird performance issue

[dwasifar]

Opnsense running on custom hardware, built on a used 8-core Xeon Supermicro server motherboard.  We'll call this device "Box" for short.

Everything has been fine, fine, fine for months and months, and then out of the blue I start seeing intermittent service dropouts.  When it's happening I can't resolve DNS, sites time out.  When I log in to Box, it is unexpectedly painfully slow to draw up; but once the dashboard is actually showing, the displayed throughput looks normal.  Navigation through Box's OPNsense interface is just generally horrible and slow.  SMART reports drive is fine.  CPU utilization has occasional unexpected spikes. 

If not for the CPU utilization, it would look like a network I/O problem.  I don't know if I/O dropouts can cause CPU spikes.  I removed Box from service and replaced it with the previous device (a converted WatchGuard) and so far everything has been stable.  I'm plugged into the same port on the LAN switch, so to my mind that eliminates the switch as the cause; this leaves the cable, or Box's network port, or some other internal Box failure (hardware or software).

My best guess is the network port on that Supermicro motherboard.  Does anyone have any other suggestions?