What is latency?
Latency is the time a packet takes to reach its destination. We explain RTT, the components of delay and why peering and shorter paths reduce it.
Last updated:
Latency is the time a packet needs to travel from sender to receiver across a network, usually measured in milliseconds (ms). In practice it is most often reported as RTT (Round-Trip Time), the time a packet takes to reach a destination and return, because that is exactly what the ping command measures. Alongside jitter and packet loss, latency is one of the three basic link-quality parameters, and it affects services that are sensitive to delay, such as gaming, VoIP, video conferencing and algorithmic trading.
What delay is made of
Total latency is the sum of several independent components that accumulate at every hop along the path. Some are dictated by physics and cannot be avoided; others depend on how the network is built and loaded. This split separates delay you can reduce from delay that is fixed for a given distance.
- Propagation delay: the time to cover the distance. In fiber the signal travels at about 200,000 km/s (roughly two-thirds of the speed of light in vacuum), which is about 5 ms per 1000 km one way.
- Transmission (serialization) delay: the time to push all the bits of a packet onto the link, set by the link bandwidth and packet size.
- Queuing delay: time spent waiting in router buffers, which grows when a link is congested (bufferbloat).
- Processing delay: the time each router spends inspecting the header and making a forwarding decision.
- Hop count: every extra router and every path detour adds its own delay.
RTT versus one-way delay
RTT measures the full round trip, whereas one-way delay covers a single direction only. The two are not always symmetric, because traffic on the internet often returns over a different path than it was sent (asymmetric routing), so RTT is not necessarily exactly twice the one-way delay. Tools such as ping and traceroute report RTT because it is easy to measure without precise clock synchronization on both ends. Accurate one-way delay measurement requires synchronized time, for example via PTP or GPS, and is defined in RFC 7679.
Why peering and shorter paths lower latency
Propagation delay is directly proportional to distance, and every hop adds queuing and processing delay, so the most effective way to cut latency is to shorten the packet's physical and topological path. Peering helps here. Instead of hauling domestic traffic through distant transit hubs in Western Europe (Frankfurt, Amsterdam), exchanging it directly at a local internet exchange (IXP) keeps it within a single country or city. Traffic between two Polish networks peering in Warsaw can have an RTT below 5 ms, while the same pair of networks exchanging traffic via transit in Frankfurt can easily exceed 30 ms. A shorter path also means fewer routers in the way, so less risk of queuing and lower jitter.
Low latency with AS202520 SkyPass
AS202520 SkyPass operates PoPs in Warsaw and Wrocław and peers at Polish internet exchanges including THINX, TPIX, WRIX and 1-IX, which keeps traffic to Polish networks local and low in latency. Broad peering shortens AS_PATH lengths and removes detours through foreign hubs. For operators without their own port in these cities we offer remote IXP access, and you can check current routes and RTT in our looking glass.
Frequently asked questions
What is a good RTT?
It depends on distance and use case. Within a single city or country an RTT below 10 ms is excellent, domestic links are typically 5-30 ms, and intercontinental traffic naturally rises to 100 ms and beyond purely because of distance.
How is latency different from bandwidth?
Bandwidth is how much data a link carries per unit of time (e.g. Gb/s), while latency is how long a single packet takes to arrive. A link can have high bandwidth and high latency at the same time, and vice versa. They are two independent parameters.
Does more bandwidth reduce latency?
Not directly. More bandwidth only reduces serialization delay and the risk of queuing under congestion. Propagation delay, the time to cover the distance, cannot be overcome with a bigger link.
How do you measure latency?
The simplest way is the ping command, which returns RTT to a given address, and traceroute, which shows the delay at each hop along the path. Measuring one-way delay precisely requires synchronized clocks on both ends.