Gigabit Router Buyer's Guide — 100M vs Gigabit Ports, Selection and Testing

01 100M versus gigabit ports: more than a number

Port speed is at heart a question of how many bytes can move per second. A 10/100Mbps port (a 100M port) caps at 100Mbps one way and delivers roughly 90–94Mbps once protocol overhead is removed; a 10/100/1000Mbps port (a gigabit port) caps at 1000Mbps and delivers around 940Mbps. A gigabit port can therefore carry the full bandwidth of any sub-gigabit line, while a 100M port pins you at ninety-odd megabits no matter how fast the Wi-Fi claims to be. The first consequence is an unused line. On a 200M, 500M or 1000M connection, a 100M WAN port still tops out in the low nineties, so the extra bandwidth is wasted — you buy a gigabit line and use less than a tenth of it. The second is an internal bottleneck. Copying from a NAS, backing up camera footage or moving data between devices all runs at LAN-port speed: on a 100M port the ceiling is ninety-odd megabits, so large files crawl; a gigabit port cuts the same transfer to roughly a tenth of the time. Many projects that complain of fast Wi-Fi but slow file copies trace it back here. The third is easily missed: LAN port count. A router usually offers only three or four LAN ports, so feeding switches, APs, a NAS and an NVR runs short, forcing a second switch and another layer of bottleneck. When choosing a gigabit router, read the WAN and LAN speed and the LAN count together rather than a single 'gigabit' badge.

02 100M, gigabit and 2.5G ports compared

03 Wired vs wireless: a 1200M wireless model is not a gigabit port

Wireless speed and port speed are two separate ledgers, and conflating them is the most common ordering mistake. A vendor's 1200M, 3000M or 9500M is an aggregate theoretical peak across bands and spatial streams, approached only in ideal conditions with matching clients, while real end-to-end throughput sits well below it. Port speed is different: it is a hard physical ceiling on how many bytes per second can move. The trap is treating the two as one. A router rated at 1200Mbps of wireless may still carry only 100M ports. Our CF-WR616AC V2 is exactly this case — 1200Mbps wireless with 10/100Mbps WAN and LAN, so however high its wireless figure, traffic out of the box runs at the ninety-odd megabit ceiling. By contrast, the equally 1200Mbps CF-N5 V2, CF-WR619AC V2 and CF-WR651AC carry gigabit ports and genuinely earn the name gigabit router. The test is simple: the wireless spec answers how fast the air can go, and the WAN/LAN spec answers how much the wired side can feed. What actually sets your browsing and internal transfer speed is usually the wired port. Before buying, ask one question: are the WAN and LAN 10/100 or 10/100/1000? That single question prevents most selection errors.

04 CPU and NAT forwarding: the second gate behind the port

The port is the ceiling, and whether you reach it also depends on the processor's NAT forwarding capacity. A router must open every packet, look it up, rewrite addresses and send it on, and that work falls to the CPU (or a hardware-accelerated forwarding engine). A gigabit port with insufficient forwarding capability will visibly drop in throughput once many connections, QoS or multiple WAN policies are in play. So do not stop at the word 'gigabit': confirm three things together. First, memory (RAM/Flash) — our gigabit-tier models generally start at 128MB or 256MB, leaving room for the connection table and concurrent sessions. Second, whether hardware NAT acceleration is supported, which steadies throughput under load. Third, whether WAN and LAN are balanced, avoiding a top-heavy pairing such as a gigabit WAN with 100M LAN. For hotels and campuses carrying several APs and many clients at once, forwarding capacity matters as much as port speed: the port sets the theoretical peak, and the forwarding engine decides how much of that peak survives under real load.

05 Scenarios that genuinely require gigabit

• Line above 100M: on a 200M, 500M or 1000M connection the WAN port must be gigabit, or the extra bandwidth goes entirely unused.
• Multi-camera backhaul: several IP cameras streaming to an NVR or cloud around the clock saturate a 100M port quickly, which is where stuttering, dropped frames and broken playback usually begin.
• NAS and large internal transfers: copying assets, backing up footage and moving files across the LAN all run at LAN-port speed; gigabit lifts the ceiling from ninety-odd to over nine hundred megabits.
• Several APs or switches aggregating: when multiple APs and switches hang off one router, their aggregated uplink easily exceeds 100M, and only a gigabit port avoids becoming the master bottleneck.

06 Gigabit router selection table: match ports to scenario

07 Acceptance: test with iperf, not the label

• Measure wired throughput with iperf3: one PC on a LAN port as server, another on the WAN side or a second PC as client, running each direction for about 60 seconds and reading the steady figure rather than the peak.
• Compare against normal ranges: about 90–94Mbps on a 100M port and around 940Mbps on a gigabit port are healthy; if much lower, check the cable (Cat5e or better, all eight wires) and the negotiated port speed first.
• Load-test with real traffic: run camera backhaul, large NAS transfers and multi-AP uplinks together to see whether a bottleneck appears, rather than signing off on an idle test.
• Record the negotiated speed and duplex of every backbone link, confirming no device has fallen back to 100M and none is running half-duplex.
• Write the WAN/LAN port specs, measured throughput and cable models into the acceptance record so operations has something to refer to.

Questions