How to use iPerf to test network throughput

iPerf3 is a command-line utility that measures throughput between two endpoints, helping determine whether a slow connection comes from Wi-Fi, local equipment, a router, or an internet service. It is especially useful for checking links involving a NAS, switch, or small office.

A sensible test starts with two wired devices on the same LAN, then repeats the measurement across the internet. Keep each stage separate: otherwise, a strong Wi-Fi result can hide a congested NBN connection, while an internet test can conceal poor Ethernet or cabling.

Install a matched iPerf3 pair

iperf3 uses a client and server. Install the same software version on both machines; a Synology NAS may support it through packages or a container, while Ubuntu, macOS, Windows, and some Linux routers can run it directly. The network testing notes provide useful background for anyone choosing equipment for a home lab.

On macOS, Homebrew can install iperf3; on Debian or Ubuntu, use sudo apt install iperf3; Windows users should obtain iperf3 from a reputable software source. Close VPNs, proxies, and heavy backup jobs because they consume bandwidth or processing time and can distort the result.

Establish a wired local baseline

Connect the server and client directly with Ethernet where possible, or through the same switch. Disable Wi-Fi on both endpoints, note each network card’s negotiated link speed, and test several file transfers as a sanity check. If a 1 Gbps link reports only about 100 Mbps, suspect a cable, port, or duplex setting before blaming the ISP.

Start the server with iperf3 -s, then run iperf3 -c 192.168.1.50 -t 30 -i 1 from the client, replacing that address with the server’s local IP. A healthy gigabit LAN should normally deliver well above 900 Mbps on modern hardware. Add -R to reverse the direction, and -P 4 when checking whether several streams perform better than one.

Measure the internet path

For an internet test, run the server on a VPS or another location outside the home network, then connect the client to it. Allow port 5201 through the server firewall only while testing. Australian NBN access commonly uses FTTP, HFC, or FTTN, and advertised plan tiers range from 50 Mbps to gigabit speeds; actual results also depend on the access node, router, and time of day.

Test both upload and download, using -R for one direction, and repeat during the evening when household traffic is high. Australian users can compare servers in Sydney, Melbourne, Brisbane, or Perth to examine ISP routing, although physical distance is only part of the story. Compare measured results with plans from Telstra, Optus, TPG, or iiNet rather than relying on the national headline speed.

Separate Wi-Fi from wired performance

If the wired result is strong, move one endpoint to Wi-Fi and test again. Dense flats in Sydney and Melbourne often contain many neighbouring 2.4 GHz networks, so use 5 GHz or 6 GHz where available, place the access point away from metal and microwaves, and compare channels. Repeating the test near the router shows whether range, rather than bandwidth, is the issue.

Check switches, routers, and NAS network ports for negotiated speeds, duplex errors, and thermal throttling. Jumbo frames can raise efficiency on a controlled LAN, but every device must support the larger MTU; read about jumbo frame support before changing it. Never assume an internet path accepts frames above 1,500 bytes.

Read and document the results

Read TCP and UDP output separately. TCP provides retransmission and congestion information, while UDP can expose jitter and packet loss when a target rate is set with -u -b. Retail NBN plans use Mbps, whereas file transfers often display MB/s, so divide bits by eight before comparing the figures. Dropping the capital B in Australian tech discussions can create an eightfold error.

Record the date, endpoint hardware, wired or wireless status, server location, direction, and any VPN or backup activity. A plain screenshot is usually enough for a home diagnosis. If results are going to colleagues or clients, a short visual summary can be easier to understand, and business video coupons can help compare production options.

Build a practical troubleshooting plan

Classify each result rather than chasing a single peak number. Low LAN throughput points to cabling, a switch port, Wi-Fi coverage, or endpoint limits; good LAN speed with poor internet speed points toward the NBN path or ISP. A server that performs well in reverse but poorly in the other direction may reveal upload congestion or a one-way fault.

Change one variable at a time and keep a dated baseline for future comparisons. The practical takeaway is to use iPerf3 first for a wired LAN test, then repeat it over the internet in both directions; preserve the results before changing hardware, cabling, Wi-Fi placement, or your Australian broadband plan.