WAN Latency Calculator

Route preset:

Preset loads: {{ activeRoutePresetNote }}

{{ activeRoutePreset.shortLabel }} Target RTT {{ params.target_rtt_ms }} ms {{ params.transit_hops }} hops

Route distance:

Enter one-way WAN core distance, for example 650 km or 400 mi.

Fiber profile:

Reference model: {{ activeFiberNote }}

{{ activeFiberProfile.label }} n = {{ params.fiber_profile === 'custom' ? params.custom_index : activeFiberProfile.index }} {{ params.fiber_profile === 'custom' ? 'Manual propagation model' : 'Pre-tuned terrestrial model' }}

Custom propagation index:

Use refractive index n, typically 1.467-1.470 for telecom fiber.

Transit hops:

Count routed, inspected, or carrier handoff stages as whole hops.

hops

Fixed device delay:

Enter total steady-state device latency in milliseconds.

Target RTT:

Set 0 to hide target-fit warnings and distance headroom.

Jitter budget:

Use 0% for a pure base estimate; typical planning cushion is 10-25%.

Queue reserve:

Enter extra milliseconds held aside for busy links or policy queues.

Return-path asymmetry:

0% keeps both directions equal; 5% means the return path is 5% longer.

Access loop per site:

Enter one endpoint loop length in km; the calculator applies it at both sites.

Route contingency:

Percent is applied to the core route before return asymmetry.

Hop penalty:

Use milliseconds per hop, such as 0.03 for fast backbone routing.

ms/hop

Access link rate:

Enter the bottleneck access speed in Mbps, or 0 to skip serialization.

Mbps

Packet size:

Use 64-9216 bytes; 1500 bytes fits standard Ethernet MTU.

bytes

Transaction round trips:

Use 1-100 request-response trips; chatty login flows may need several.

RTT

Budget line	Value	Copy
{{ row.label }}	{{ row.value }}

Area	Signal	Detail
Primary driver	{{ result.guidance.driverTitle }}	{{ result.guidance.driverDetail }}
Best lever	{{ result.guidance.nextActionTitle }}	{{ result.guidance.nextAction }}
Measure next	{{ result.guidance.measureTitle }}	{{ result.guidance.measureNext }}

Delay driver	Share / value	Copy
{{ row.label }}	{{ row.value }}

Lever	Budgeted RTT gain	Configured flow gain	Why it helps	Copy
{{ row.label }}	{{ row.rttGainDisplay }}	{{ row.flowGainDisplay }}	{{ row.note }}

Distance-budget line	Value	Copy
{{ row.label }}	{{ row.value }}

Lane	What changes	Budgeted WAN RTT	Configured flow	Gain vs current	Flow fit	Copy
{{ row.label }} {{ row.note }}	{{ row.changeSummary }}	{{ row.rttDisplay }}	{{ row.flowDisplay }}	RTT {{ row.rttGainDisplay }} Flow {{ row.flowGainDisplay }}	{{ row.flowFitText }}

Pattern	Round trips	Base floor	Budgeted floor	Flow budget fit	Copy
{{ row.label }} {{ row.note }}	{{ row.roundTripsLabel }}	{{ row.baseDisplay }}	{{ row.budgetedDisplay }}	{{ row.fitText }}

Operational window	Published RTT window	Current fit	Why it matters	Copy
{{ row.label }} {{ row.source }} \| {{ row.sourceHint }}	{{ row.windowDisplay }}	{{ row.statusText }} {{ row.gapDisplay }}	{{ row.note }}

Field	Value	Copy
{{ row.label }}	{{ row.value }}

Embed:

Customize

Include current inputs

Size

Advanced

Width

Height

Aspect ratio

Max height

Collapsible embed

Allow fullscreen

Referrer policy

Sandbox tokens

Bandwidth and latency answer different network questions. Bandwidth says how much data can move through a path at once. Wide area network latency says how long a request has to wait for the path to carry a packet out and bring the reply back. That wait often sets the feel of remote desktop, storage, voice, database, and branch-office applications long before a link is full.

Distance creates the first floor. Electromagnetic signals have a fixed speed in vacuum, and light in telecom fiber is slower because the glass has a refractive index above 1. Carrier routes also follow conduits, rights-of-way, metro rings, cable landing stations, peering points, and access tails, so the path distance is usually longer than a straight map line. A nearby regional path can stay comfortable for interactive work, while an intercontinental path may be acceptable for bulk transfer but tiring for software that waits after every request.

RTT: Round-trip time, the send-and-return delay commonly reported by ping and many monitoring tools.
Propagation delay: The distance floor caused by signal speed through the route medium.
Serialization delay: The time needed to clock packet bits onto the bottleneck access link.
Jitter: Variation in packet delay, often caused by queues, contention, shaping, or changing paths.

Network teams usually care about the single round trip and the number of request-response turns in the application. One 70 ms RTT may be acceptable for a simple request. The same path becomes painful when login, authentication, file metadata, API calls, or storage acknowledgements require several serial turns before the user sees progress. Latency planning belongs beside application design because fewer round trips can help when the geography cannot change.

WAN route diagram showing propagation, hops, device and queue delay, jitter reserve, and a longer return path — A WAN RTT budget combines the fiber distance floor with route overhead, fixed device delay, queue reserve, serialization, and margin for delay variation.

A useful latency estimate separates fixed physics from choices a team can still influence. Geography, carrier routing, and fiber properties set a lower bound. Hop count, middleboxes, queues, access speed, and packet size add smaller but still important terms. Application round trips multiply the user-visible wait. Planning is strongest when those parts are visible before a team moves workloads, changes a carrier path, tunes queues, or redesigns a chatty workflow.

How to Use This Tool:

Start with the route shape, then add the overhead that sits on top of propagation. Use the target only after the distance, fiber, hop, queue, and workflow assumptions look realistic.

Choose a Route preset such as metro, regional, cross-country, or transoceanic, or keep Custom path when measured assumptions are ready.
Enter the one-way Route distance and distance unit. Use carrier route length or a conservative route estimate rather than straight-line map distance.
Select the Fiber profile. Use Custom propagation index only when you have a specific refractive-index assumption.
Set Transit hops, Fixed device delay, and Target RTT. A target of 0 turns off target-fit and distance-headroom comparisons.
Open Advanced for jitter budget, queue reserve, return-path asymmetry, access loops, route contingency, hop penalty, access link rate, packet size, and transaction round trips.
Read the summary first, then use Latency Budget, WAN Path Guidance, and WAN Delay Drivers to identify the main delay source.
Use RTT Improvement Levers, Route Distance Budget, WAN Scenario Lanes, and Workload RTT Windows to compare path cleanup, queue reduction, and fewer application turns.

If a named preset changes to Custom path after edits, the current values no longer match that preset. If an input refuses a value, check the visible bounds, especially packet size, custom propagation index, jitter budget, hop penalty, and transaction round trips.

Interpreting Results:

Budgeted WAN RTT is the planning number because it includes the jitter reserve. Base RTT before jitter is the lower estimate after distance, hops, fixed device delay, queue reserve, and serialization are counted. The propagation share shows whether geography dominates the path, while the device, queue, hop, and serialization rows show delay that may be reducible without moving sites.

Latency envelope bands used for WAN latency interpretation
Budgeted WAN RTT	Envelope	Typical read
< 20 ms	Metro	Good fit for many latency-sensitive interactive paths.
20 to < 60 ms	Regional	Often usable for regional applications, real-time media, and remote access when loss and jitter stay low.
60 to < 120 ms	Long haul	Application round trips, remote desktop feel, and media quality need closer review.
≥ 120 ms	Intercontinental	Regional placement, caching, and fewer request turns usually matter more than extra bandwidth.

Target fit is measured against the budgeted RTT. At least 15 ms of spare target headroom is treated as a comfortable within-target result. A non-negative gap below 15 ms is still inside the target but narrow. A miss under 15 ms is close enough that route cleanup or queue changes may help, while a miss of 15 ms or more usually needs a shorter path, fewer request turns, or a higher target.

Do not treat a green target result as proof that users will be satisfied. Packet loss, Wi-Fi, server processing, Domain Name System lookups, Transport Layer Security setup, storage waits, and burst queues can still dominate live experience. Compare the modeled floor with p50 and p95 RTT, traceroute from both ends when possible, and real transaction timing for the application path.

Technical Details:

WAN latency has a distance-driven propagation term and several operational terms. Propagation delay depends on the total forward and return path length divided by signal speed in fiber. Operational delay adds forwarding, inspection, queueing, and serialization time. A jitter reserve then raises the base estimate so the result behaves more like a planning budget than an ideal idle-path measurement.

The model uses the exact vacuum speed of light, converts it through the selected fiber propagation index, and then builds a forward path from the core route distance, route contingency, and two access loops. Return-path asymmetry lengthens only the reverse leg. Hop delay, fixed device delay, queue reserve, and access-link serialization are added before the jitter uplift is applied.

Formula Core:

Signal speed in fiber is the vacuum speed of light divided by the active propagation index.

v = \frac{c}{n}

D_{forward} = D_{core} \times (1 + r) + 2 \times D_{loop}

D_{return} = D_{forward} \times (1 + a)

{RTT}_{prop} = 1000 \times \frac{D_{forward} + D_{return}}{v}

{RTT}_{serial} = 2 \times \frac{8 \times P}{B \times 1000000} \times 1000

{RTT}_{budget} = ({RTT}_{prop} + H \times L_{hop} + E + Q + {RTT}_{serial}) \times (1 + j)

WAN latency formula symbols and units
Symbol	Meaning	Unit or source
`c`	Speed of light in vacuum	299,792.458 km/s
`n`	Fiber propagation index	Preset or custom index
`D_core`, `D_loop`	Core route distance and access loop per site	km after unit conversion
`r`, `a`, `j`	Route contingency, return-path asymmetry, and jitter reserve	Percent converted to decimal factors
`H`, `L_hop`	Transit hops and per-hop delay	hops and ms/hop
`E`, `Q`	Fixed device delay and queue reserve	ms
`P`, `B`	Packet size and access link rate	bytes and Mbps

With the cross-country default, a 4,200 km core route plus 10 percent contingency and 6 km access loops per site becomes a 4,632 km forward path. With n = 1.468, the effective speed is about 204,218 km/s, so symmetric propagation contributes about 45.363 ms RTT. Adding 14 hops at 0.05 ms each, 2.4 ms of fixed device delay, 3 ms of queue reserve, and 0.024 ms of access-link serialization gives about 51.487 ms before jitter. A 20 percent jitter reserve raises the planning number to about 61.785 ms.

WAN latency input bounds that affect repeatable comparisons
Input	Range or behavior	Why it matters
Custom propagation index	1 to 2	Changing n changes the physical speed used for the propagation floor.
Jitter budget	0% to 200%	Higher reserve widens the gap between idle estimates and planning RTT.
Return-path asymmetry	0% to 100%	Only the reverse leg is lengthened, so RTT rises without changing the forward route label.
Access link rate	0 Mbps or higher	0 removes serialization; positive values model packet clocking on the slower edge link.
Packet size	64 to 9,216 bytes	Packet size affects serialization delay, especially on slower access links.
Transaction round trips	1 to 100 RTT	Each serial request-response turn adds one budgeted WAN RTT to the workflow floor.

Accuracy Notes:

The estimate is a planning model, not a live measurement. It does not probe the route, contact endpoints, or verify carrier path length.
The calculation runs in the browser after the page loads, so entered route assumptions are not sent to a remote measurement service by this calculator.
Carrier routes can change, especially during failures or maintenance, so compare the estimate with production p50 and p95 samples before committing to an SLA.
Loss, reordering, congestion control, application server time, encryption handshakes, name lookups, and storage waits can make user experience worse than RTT alone suggests.
Published workload windows are planning references. They do not guarantee quality when packet loss, jitter, device load, or endpoint performance is poor.

Worked Examples:

WAN latency worked examples
Case	Inputs	Result cue	Follow-up
Cross-country backbone	4,200 km core route, n = 1.468, 14 hops, 20% jitter reserve, 80 ms target, 6 RTT workflow.	Budgeted WAN RTT is about 61.8 ms, leaving roughly 18.2 ms target headroom. The configured workflow floor is about 371 ms.	Latency is inside the target, but a 6-turn flow still makes every page or transaction wait on several WAN crossings.
Regional WAN with a narrow target	650 km route, 7% contingency, 8 hops, 12% jitter reserve, 25 ms target.	Budgeted RTT is about 11.3 ms, but the target margin is below 15 ms, so the fit is narrow rather than generous.	Watch p95 RTT and queue reserve. A small detour or burst queue can erase the spare margin.
Distance budget says no path room	Very low target with high queue reserve, many hops, or heavy fixed device delay.	The route distance budget can report that fixed delay already exhausts the target before propagation is counted.	Lower queue reserve, hop count, or device delay first. Shortening the route cannot fix a target already spent on non-distance delay.

FAQ:

Why is the RTT higher than the fiber distance alone?

Fiber distance gives the propagation floor. The estimate also includes route contingency, access loops, return-path asymmetry, hop delay, fixed device delay, queue reserve, serialization, and jitter margin.

Can more bandwidth fix high WAN latency?

More bandwidth can reduce serialization delay and congestion risk, especially on slow access links. It cannot remove the propagation floor caused by distance and signal speed.

Why did my preset become Custom path?

Named presets represent fixed scenario assumptions. Once you edit fields such as distance, hops, queue reserve, or target RTT, the current input set no longer matches the preset exactly.

Why do workload windows differ from my target?

Your target is a local planning goal. Workload windows compare the modeled RTT with published reference points for specific media, desktop, or collaboration experiences.

Does this measure my actual WAN path?

No. It estimates a path from the values you enter. Use live RTT, traceroute, device telemetry, and application timing to verify the model against real traffic.

Glossary:

Core route distance: The one-way WAN path distance before route contingency and access loops are added.
Route contingency: A percentage reserve for route detours, rights-of-way, landing stations, or uncertain cable length.
Access loop: The local tail distance from each endpoint to the WAN core or carrier handoff.
Hop penalty: The modeled per-hop forwarding delay for routers, firewalls, peering handoffs, or inspected stages.
Distance headroom: The remaining core route distance that can fit inside a target RTT after fixed delays and jitter reserve are counted.