Industrial Network Redundancy: Multiple Ports, Link Backup and Failover
Downtime is expensive on an industrial site, and networking is the most failure-prone part of the chain. The goal of redundancy is not “no failures” but “a failure that does not stop production”.
Physical separation beats VLANs
Using VLANs to separate traffic by purpose is standard practice, but on industrial sites physical separation is often preferred — it is more intuitive, easier to get signed off, and does not depend on a switch being configured correctly.
The 03R EDGE-RT28 industrial router board integrates 3× gigabit LAN plus 1× gigabit WAN, all configurable. Three LAN ports allow physical separation of the internal device segment, the management segment and the external segment — closer to on-site practice than a single port with VLANs.
Link backup
The common pattern is wired primary, cellular secondary: switch to 4G/5G when the primary link drops and back when it recovers. EDGE-RT28 provides a Mini PCIe slot for a cellular module, achievable with policy routing under OpenWrt. What matters is the acceptable failover time — some scenarios need seconds, others tolerate minutes, and that directly determines solution complexity.
The device itself needs redundancy too
Backing up the link while leaving the device as a single point of failure does not improve overall reliability much. Where availability matters, plan for hot standby or at least spares that can be swapped quickly. That also means paying attention to lead times and long-term availability at selection time.
On the compute side, the 05A IPA-RK3588 industrial control board offers 2× gigabit Ethernet, 4× RS-232/485 (isolated) and 1× CAN-FD, with -20℃ to +70℃ operation and an MTBF of ≥ 50,000 hours. Where a site needs both networking and local compute, it fits better than a general-purpose development board.
Do not neglect observability
Whether a redundant link ever actually switched, how many seconds it took and why — none of that can be answered without logs. Record link state, failover events and traffic anomalies, and review them periodically. Many redundancy schemes only reveal configuration problems when a real failure occurs.
Specifications are on the EDGE-RT28 product page.
