Check whether a proposed network path fits inside an application's one-way latency target. Enter the known or estimated delay from propagation, routing, queueing, endpoint processing, and a safety margin. The calculator totals those components, shows the remaining budget, and identifies the largest contributor. It is designed for architecture reviews and troubleshooting plans rather than replacing packet captures or active path measurements.
A positive result is unallocated time inside the target. A negative result is the amount by which the proposed path exceeds the target.
Round-trip time is not always exactly twice one-way delay because routes and queue conditions can differ in each direction.
A 100 ms target minus 20 ms propagation, 12 ms routing, 18 ms queueing, 15 ms processing, and 10 ms margin leaves 25 ms remaining. Queueing is the largest controllable component.
No. A budget allocates an allowed target among components; measurement confirms what the deployed path actually delivers.
Convert it to an appropriate one-way target first. Simply halving RTT is an estimate because the two directions may differ.
The margin absorbs normal variation, measurement uncertainty, and conditions not represented by the component estimates.
Insufficient capacity can increase queueing delay, but raw link bandwidth is not itself a latency component.
Start with the largest controllable contributor, often queueing or endpoint processing, while confirming every estimate with measurements.
| Component | Typical source |
|---|---|
| Propagation | Physical path distance and medium |
| Routing/switching | Network-device forwarding |
| Queueing | Traffic waiting behind other packets |
| Processing | Application, encryption, and endpoint handling |