Latency

Latency is the round-trip delay a packet experiences travelling from your computer to a server and back, measured in milliseconds (ms). It is not the same as bandwidth, which is how much data can move per second. A link can have huge bandwidth yet still feel sluggish when latency is high, because every request has to wait for that round trip to complete. The quickest way to sample latency is a ping, while the route packets actually take can be inspected hop by hop with traceroute. Physical distance, congestion and each intermediate device all add to the figure.

Routing, congestion and the target affect latency and throughput. Compare the same workload through each relevant path at several times instead of assuming predictable performance from a region. s4m has selectable public exit configurations. Use the endpoint or profile issued for the protocol and location selected in your account. Related topics: proxy, VPN.

To cut latency in practice, keep the number of hops small and the exit close to your target, reuse a persistent connection so you avoid repeated handshakes, and resolve DNS nearby. When measuring, average several samples and watch jitter alongside the mean rather than trusting one figure. The s4m tools can help you check the round trip your traffic actually takes through an exit before you commit a latency-sensitive job to it.