Fiber Latency per Kilometer: A Route Budget Worksheet

Summary: A fiber optic latency per km calculation estimates propagation delay from optical path length and group index. For conventional silica single-mode fiber, about five microseconds per kilometre is a useful first planning approximation. Build the complete service budget separately from equipment processing, packet serialization, queuing and the return route.

The calculations below use a hypothetical group index of 1.468 and the vacuum speed of light, 299,792,458 metres per second. They are design examples, not measurements of Liqiba cable or a promise of application performance.

Define the latency boundary first

A cable propagation figure describes time spent travelling through fiber. A service latency measurement may start at an Ethernet port, a server application or a user’s device. Those boundaries include different components, so two numbers cannot be compared until their start and end points are defined.

The Fibre Channel Industry Association’s data-center interconnect discussion uses approximately five microseconds per kilometre for fiber transport and distinguishes that contribution from latency within the data center. This is a useful separation when evaluating a route proposal.

For procurement, specify whether the target is one-way delay, round-trip delay, a percentile under traffic or a maximum under a defined operating condition. “Low latency” without those details gives the cable supplier and network designer no common acceptance target.

Calculate propagation using group index

Use t = n_g × L ÷ c, where n_g is the relevant group index, L is optical fiber length in metres and c is the vacuum speed of light. Convert seconds to microseconds by multiplying by one million.

For one kilometre and the assumed index, t = 1.468 × 1000 ÷ 299,792,458, approximately 4.90 microseconds. The five-microsecond shortcut is slightly conservative relative to this particular assumption. For a contractual limit, use the specified fiber data and operating wavelength rather than silently treating the shortcut as a measured property.

Group index is the appropriate propagation quantity for the signal envelope. Do not substitute an arbitrary refractive-index number found in a glass description. The site’s OTDR group-index guide also explains how the chosen index affects inferred optical length.

Use installed optical length instead of map distance

A straight line between two buildings is normally shorter than the cable route. Duct detours, road crossings, vertical risers, access loops and repair slack add length. Fiber inside a cable can also be longer than the jacket length because of the cable construction.

Prepare a length hierarchy: geographic separation, planned pathway length, installed cable length and estimated optical fiber length. Record which number each calculation uses. If an OTDR supplies the length, record its index setting and the launch-lead boundary so the resulting estimate remains interpretable.

FOA’s OTDR testing guide explains that the fiber can be longer than the cable and that the instrument’s refractive-index setting affects length measurement. Keep these construction and measurement effects visible rather than equating a jacket marking with exact optical path length.

Compare a primary route and a diverse route

Assume a primary path has 24 km of estimated optical fiber and a physically diverse backup has 31 km. Both use the same hypothetical group index. The calculated propagation values are rounded planning figures.

The backup path adds propagation delay during failover even if bandwidth and optics remain unchanged. Review whether the application target must hold on both routes. Do not shorten a route merely to improve the number if doing so removes the physical diversity that the resilience design requires.

Handle asymmetric return paths

If traffic travels outward on the 24 km route and returns on the 31 km route, propagation RTT is approximately 117.5 + 151.8 = 269.3 microseconds. It is not twice either individual value. Direction-specific routing and protection behavior therefore belong in the latency worksheet.

Halving an observed round-trip time gives only an estimated one-way value when symmetry is justified. Accurate one-way measurement requires suitable synchronized test endpoints and a defined timestamping method. Clock error can dominate a small delay difference, so document synchronization performance with the result.

Add the equipment and packet terms explicitly

Build a table of the actual endpoints, switches, transponders, encryption devices and other functions in the service path. Obtain processing-delay evidence for the selected operating mode, including relevant FEC or packet-handling behavior. Do not copy a latency number from another firmware or line-rate configuration.

Serialization is the time to place a frame’s bits onto an interface. As a simplified calculation, 1500 bytes contain 12,000 bits, requiring 1.2 microseconds at 10 Gb/s and 0.12 microseconds at 100 Gb/s. Those figures exclude Ethernet framing, preamble, interpacket gap and any other protocol overhead.

Higher bitrate reduces that serialization term; it does not make light propagate faster through the same installed fiber. Likewise, queue delay depends on traffic and scheduling. An unloaded test cannot demonstrate a loaded percentile target unless the agreed test profile covers that condition.

Allocate a budget without double-counting

Suppose a hypothetical service has a 250-microsecond one-way target. Allocate 117.5 microseconds to primary-route propagation, 60 to documented equipment and packet handling, and 40 to the permitted traffic-dependent allowance. The total is 217.5 microseconds, leaving 32.5 for the remaining design reserve.

Substituting the backup route raises the total to 251.8 microseconds under the same assumptions, exceeding the target by 1.8. This is a planning warning, not a measured service failure. It identifies the need to review route length, equipment assumptions or the target before procurement.

Check the scope of equipment figures before adding them. A published port-to-port delay may already include a function listed elsewhere in the worksheet. Mark inclusions and exclusions beside every term to prevent counting the same processing or serialization delay twice.

Validate the estimate during commissioning

Measure the actual service between the agreed boundaries using the specified packet sizes, load, duration and direction. Save the device configuration and active route with the timestamps. Repeat the relevant check after a controlled protection switch if the target applies to backup operation.

Use discrepancies to investigate assumptions rather than editing the calculated fiber length until it matches the result. Additional switching, a different return path, queueing or test-clock error can explain excess delay. A route-length calculation remains a propagation estimate, even when it happens to match one observed packet result.

Test how sensitive the estimate is to length

Using the same assumed index, an extra 100 metres adds approximately 0.49 microseconds one way. An unplanned two-kilometre detour adds about 9.79 microseconds. These calculated increments help decide which survey uncertainties matter to a particular service target.

Do not remove necessary installation or repair slack simply because it contributes a small delay. Instead, assign realistic slack at design time and include it in the optical-length estimate. The installation team can then preserve bend radius and maintenance access without unexpectedly changing the budget.

For a length estimate expressed as a range, calculate both endpoints. If the optical path is expected to be between 24 and 25 km, propagation lies approximately between 117.5 and 122.4 microseconds under these assumptions. Carry that range into the service budget rather than reporting an apparently exact value from an uncertain route drawing. Later, replace the estimate with the as-built evidence and retain the earlier assumptions for comparison.

Give suppliers a route-based requirement

The IPoDWDM versus transponder planning guide helps define the equipment architecture. For the passive route, use the fiber optic cable supplier page and provide installed-length expectations, fiber category, wavelength, environmental requirements and the information needed to support propagation estimates.

Ask separately for optical attenuation data and the applicable group-index information. Keep construction slack, route reserve and future repair allowances visible in the design. A transparent budget lets an owner evaluate the trade-off between shorter delay, practical installation and a resilient route without treating bandwidth or low loss as substitutes for measured service latency.