Loverpinger.

Performance

Why bandwidth does not tell the whole story.

A high transfer rate does not guarantee a responsive application.

Throughput and waiting are different.

Throughput describes how much data can move over time. Latency describes how long an exchange takes. A small interactive request can spend much of its time waiting, even on a connection capable of transferring large files quickly.

Dependencies multiply round trips.

If one request must finish before the next begins, each exchange adds another wait. Four sequential exchanges over a 40 ms round-trip path contribute roughly 160 ms of network waiting, before server processing and data transfer are counted.

Loss changes the experience.

Missing packets can trigger recovery and change transport behaviour. Occasional bursts may matter more to an interactive application than an average transfer rate suggests. Record the distribution of delays and correlate problems with the time users report them.

Practical takeaway

Measure representative actions. Use bandwidth tests as one input alongside latency, loss and application timings.

← All insights