VPS Speed Testing: speedtest, iperf3 and Real Download Tests

"How fast is this VPS, really?" Gut feeling is unreliable — measure it. This article covers three methods from quick to precise, and how to read the results correctly.

Method 1: speedtest-cli (quickest)

Install the speedtest CLI on the VPS and run it; it picks a nearby server and measures up/down bandwidth. Caveat: this measures VPS-to-speedtest-server throughput, which is usually fast — it is not the same as the speed from your own location to the VPS.

Method 2: real download test (closest to user experience)

Place a test file in the VPS web root (e.g. a 100 MB file generated with dd), then download it from your local machine and watch the actual speed. This is the exact path your users take, so it reflects real experience best. Download speed (MB/s) × 8 ≈ bandwidth (Mbps).

Method 3: iperf3 (precise point-to-point)

iperf3 streams traffic directly between two machines: run the server on the VPS and connect with a local client. It measures upload and download separately and can saturate the link with parallel streams — ideal for questions like "the plan says 100M, why am I only getting 20M?".

Three keys to reading results

• Test at different hours: international routes slow down at evening peak; one measurement proves nothing — test morning, afternoon and night;
• Separate bandwidth from latency: enough bandwidth with high latency means fast large downloads but slow page loads. Check latency with ping and bandwidth with downloads — measure both;
• A slow single stream does not mean a bad route: cross-border single-stream TCP is sensitive to packet loss. If parallel streams fill the link, raw bandwidth is fine — tune instead (e.g. enable BBR congestion control).

If measured numbers fall far short of the plan's rating, send your test time, method and results to 00Shark support and we will determine whether it is route fluctuation or a configuration issue.