Upgraded a structured-cabling apartment from 2.5Gb/s to 10Gb/s using MikroTik switches, DAC cables, and SFP+ cards, with iperf3 confirming near-line-rate throughput.
Key Takeaways
Unknown wall cabling (possibly CAT-5E/6/6A) handled 10Gb/s on short runs; iperf3 showed ~9.9Gb/s one direction between a Ubiquiti USB dongle and an Asus XG-C100F PCIe card.
USB-attached 10G dongles bottleneck on single-core IRQ handling; a ThinkPad maxed out at 7Gb/s receive due to ksoftirqd saturation.
MikroTik CRS305-1G-4S+IN (study) and CRS304-4XG-IN (patch panel, living room) are the core switches; DAC cables keep heat lower than active SFP+ transceivers.
Thermals are the persistent problem: 10GBASE-T SFP+ modules in unventilated sideboards and cupboards run above 70C internal; Protectli VP2420 case acts as a passive heatsink.
Grafana + InfluxDB + Telegraf + SNMP scraping from reggie provides continuous thermal monitoring across all MikroTik switches.
Hacker News Comment Review
Commenters flagged a critical hardware split: old-gen 10G SFP+ modules (30m rated, ~3W) cause link flaps from heat; new-gen (80/100m rated, ~1.5W) run much cooler and are the correct buy.
MikroTik cable-distance limits add another variable: MikroTik S+R10J modules are rated to ~30m for 10Gbps regardless of cable category, a detail the article did not cover.
Practical consensus is that 10G “just works” on mystery cabling in many homes, and the main regret is cost, not complexity.
Notable Comments
@xxpor: Two SFP+ transceiver generations exist; new 80/100m ~1.5W modules avoid the thermal link-flap issues common with old 30m ~3W units.
@cyberax: Adding a Noctua fan or passive radiators to the MikroTik resolves thermals; heat is a density problem, not total wattage.
