| Scenario | Medium | Distance | Bandwidth | Hops | Estimated RTT | Class |
|---|---|---|---|---|---|---|
| Metro fiber link | Fiber (typical) | 50 km | 1 Mbps | 8 | 3.5 ms | Excellent |
| Cross-country fiber | Fiber (typical) | 3000 km | 2 Mbps | 12 | 38 ms | Good |
| Wireless last‑mile | Wireless/Air | 10 km | 5 Mbps | 6 | 29.1 ms | Good |
| GEO satellite hop | GEO Satellite (approx.) | 2000 km | 25 Mbps | 4 | 525.3 ms | High |
- Enter the path distance and choose a unit.
- Select a medium preset or provide a custom speed.
- Set bandwidth and typical request/response payload sizes.
- Add hop count and per-hop processing, if known.
- Optionally enter utilization for queueing estimation.
- Click Calculate to view results above.
- Measured latency can differ due to route changes and buffers.
- Wireless and satellite often add scheduling overhead.
- For satellites, the calculator adds typical altitude path.
- Use the breakdown to see what dominates your RTT.
Propagation Sets the Floor
Even with perfect hardware, light travel time dominates long paths. Fiber carries signals at roughly 200,000 km/s, so 1,000 km adds about 5 ms one way and 100 km adds about 0.5 ms.
Serialization Depends on Bandwidth and Size
Transmission delay is packet bits divided by link rate. A 1,500‑byte frame on 10 Mbps takes about 1.2 ms to serialize, while on 1 Gbps it is near 0.012 ms. A 64‑byte request on 100 Mbps is about 0.005 ms, but a 1 MB response on 50 Mbps can exceed 160 ms.
Hops Add Processing and Buffering
Each router or switch introduces processing time and may queue traffic. Ten hops at 0.2 ms per hop contribute 2 ms per direction before queueing. Internet paths of 12–20 hops are typical; inside a datacenter, 4–6 hops is common. Middleboxes, encryption, and radio schedulers often behave like additional hops, so modeling them as fixed extra delay keeps estimates realistic.
Utilization Drives Queuing and Jitter
When utilization ρ rises, buffers fill and delay becomes volatile. In simple queue models, the factor ρ/(1−ρ) explodes as ρ approaches 1: at 70% it is 2.33, at 90% it is 9.00. That turns microseconds of serialization into milliseconds of waiting. The calculator’s jitter estimate is based on the queue component, helping compare quiet periods and busy-hour performance.
Satellite Paths Change the Geometry
Satellite RTT is shaped by altitude and gateway placement. A GEO relay adds tens of thousands of kilometers of space path, so propagation alone can exceed 500 ms RTT in practice. LEO systems shorten the path dramatically, often adding about 1,200 km up and down, but still include scheduling, inter-satellite routing, and terrestrial backhaul delays.
Interpreting Results for Real Networks
Use the breakdown to identify the dominant term, then optimize that layer. If BDP is large, increase TCP window or enable window scaling so throughput is not window‑limited, especially for file transfers and video. If queueing dominates, reduce congestion with QoS, shaping, active queue management, or capacity upgrades. Validate estimates by measuring ping RTT and tracing hop count to tune inputs and set SLO targets.
What is the difference between one-way delay and RTT?
One-way delay is the time from sender to receiver in a single direction. RTT includes the forward path plus the return path, so it is the most common metric for ping and interactive applications.
How do I estimate hop count if I don’t know it?
Run a traceroute to the destination and count the visible routers, then round up slightly for hidden hops. For private networks, count switching layers, firewalls, and WAN edges as hops too.
Why does utilization change latency so much?
As links approach saturation, packets wait in buffers behind other packets. Queueing grows nonlinearly with utilization, so small traffic increases near peak hours can add large latency and jitter.
How should I use the bandwidth-delay product result?
BDP approximates how much data must be in flight to fully use the link. If your TCP window is smaller than BDP, throughput can be limited even when bandwidth is available.
Are satellite values in the calculator exact?
They are directional estimates. The satellite modes add typical altitude path length, but real RTT also depends on ground station placement, routing, and scheduling. Use measured RTT to adjust extra delay until the model matches reality.
Does packet size include protocol overhead?
Enter payload sizes that reflect your application messages. If you want closer wire-level timing, add headroom for headers, TLS records, or encapsulation, especially on small packets where overhead is proportionally larger.