{{ summaryTitle }} {{ summaryValue }} {{ summaryLine }} {{ badge.label }}{{ badge.value }}

MetricValuePlanning meaningCopy
{{ row.metric }}{{ row.value }}{{ row.note }}
Additional failuresPaths leftEffective capacitySpare after demandStateCopy
{{ row.failures }}{{ row.paths }}{{ row.capacity }}{{ row.spare }}{{ row.status }}
ReviewSignalNext actionCopy
{{ row.review }}{{ row.signal }}{{ row.action }}
{{ formulaText }}
LACP Bundle Capacity Calculator Calculate usable LACP bundle capacity after failed links, reserve, and hash-efficiency discounts while checking aggregate and single-flow headroom.
BGP AS Path Policy Checker Audit pasted BGP route paths for policy conflicts and trace each pass, watch or reject verdict before changing production router filters.
QoS Bandwidth Allocation Calculator Allocate a QoS parent rate across traffic classes and check reserved bandwidth, best-effort headroom, burst ceilings and priority pressure.
Ethernet Frame Efficiency Calculator Calculate Ethernet payload efficiency and goodput from frame size plus VLAN and wire-time overhead, then check whether a target rate fits.

Equal-cost multipath (ECMP) routing can install several next hops for the same destination. That gives the forwarding plane more aggregate capacity and resilience, but it does not turn a group of links into one larger link for every conversation. A flow hash normally keeps each flow on one member path, so the traffic mix determines how evenly the members fill.

A capacity plan needs three different views of the same route set:

  • Physical capacity counts the healthy members and their per-path rate.
  • Effective capacity discounts that total for imperfect flow distribution.
  • Failure runway asks how many more members can disappear before demand or the selected reserve target no longer fits.

Hashing is most favorable when there are many similarly sized flows and the chosen header fields provide good entropy. A small flow count, a few elephant flows, polarization across several ECMP stages, or unequal member performance can leave one path congested while others have room. Measured path utilization is therefore better evidence for hash efficiency than a generic percentage.

The largest-flow check addresses a separate limit. One ordinary hashed flow cannot use the sum of all member rates. An 8 × 100 Gbps group may carry hundreds of gigabits in aggregate, yet a 120 Gbps conversation still exceeds a 100 Gbps member unless the application or transport divides it into multiple independently hashed flows.

ECMP capacity is a planning estimate, not a forwarding guarantee. Encapsulation, device-specific hash fields, member oversubscription, packet overhead, traffic bursts, and control-plane convergence can all change observed throughput. Compare sustained demand and link rates on the same payload basis, then verify the plan with routing state and telemetry.

How to Use This Tool:

Model the route set that is forwarding now, then test the next failure or maintenance state.

  1. Enter Total ECMP paths and Healthy paths from current routing state. Healthy paths cannot exceed the configured total.
  2. Set Per-path capacity and Aggregate demand using the same sustained-rate basis. Do not mix line rate with payload demand unless overhead is intentionally covered elsewhere.
  3. Choose Hash efficiency and Target headroom. Prefer a value derived from member telemetry; 70% to 90% is only a conservative planning range when distribution is unknown.
  4. Enter Largest single flow when a small number of large conversations matter. Use 0 only when that risk is intentionally left unmodelled.
  5. Add Planned path removals for a drain or maintenance event, then check effective capacity, target spare, single-flow margin, and the failure curve before proceeding.

Interpreting Results:

ResultMeaningDecision
Target metEffective capacity is greater than or equal to demand plus the selected reserve.Check failure tolerance and the largest-flow margin before accepting maintenance risk.
Demand fits, reserve thinEffective capacity covers demand but is below the reserve target.Review bursts, growth, and member-loss exposure.
Capacity shortfallEffective capacity is below demand.Restore paths, reduce load, increase member rate, or improve measured distribution.
Single flow tightThe entered flow is above 85% and no greater than 100% of one path.Leave operating margin or split the workload.
Single flow over pathThe entered flow is greater than one member's capacity.Aggregate capacity does not solve this flow-level limit.

A zero-demand input can report the reserve target as met even with no healthy paths because both demand thresholds are zero. That boundary is mathematically consistent but does not describe a usable forwarding service.

Technical Details:

The model treats healthy members as equal-rate paths and applies one user-supplied efficiency factor to represent uneven flow placement. It does not simulate individual hashes or packet scheduling. This makes the arithmetic useful for capacity envelopes while keeping the distribution assumption visible.

Formula Core:

Effective capacity is the healthy aggregate multiplied by hash efficiency.

Ceff=Nhealthy×Cpath×E100

The reserve target and spare values follow:

Dtarget=D×(1+H100)Starget=Ceff−Dtarget
SymbolMeaningUnit
NhealthyHealthy installed next hopscount
CpathSustained capacity of one memberGbps
EHash efficiency%
DAggregate demandGbps
HTarget headroom%

For 8 healthy 100 Gbps paths at 85% efficiency, effective capacity is 680 Gbps. Demand of 520 Gbps with 25% headroom sets a 650 Gbps target, leaving 30 Gbps above the target. One additional path failure leaves 595 Gbps, so raw demand still fits but the target no longer does.

Rule Core:

Failure tolerance tests every whole number of additional failed paths from 0 through the current healthy count. A failure count is retained when the remaining effective capacity is greater than or equal to the raw demand or target demand. Planned removals use the same capacity equation but report the resulting spare directly.

  • Capacity is Target met at equality with target demand.
  • It is Demand fits, reserve thin when capacity is at least raw demand but below target demand.
  • The single-flow status is Tight only when flow is greater than 85% of member capacity; equality at 85% still fits.
  • The single-flow status is Over one path only when flow is greater than member capacity; equality still fits.

Accuracy Notes:

Hash efficiency is a planning allowance, not a property guaranteed by ECMP. Validate it from per-member counters over representative busy periods. Also confirm the actual hash fields, symmetry requirements, encapsulation behavior, link overhead, and downstream bottlenecks on the deployed platform. All entered data stays in the browser; the page does not send a route table or traffic measurements to a server.

Worked Examples:

Aggregate target passes while one flow fails

Six healthy 100 Gbps paths at 80% efficiency provide 480 Gbps. With 400 Gbps demand and a 20% reserve, the aggregate target is met exactly. A planned one-path removal leaves 400 Gbps, enough only for raw demand, while a 120 Gbps single flow is 20 Gbps above one member. The maintenance event has no spare capacity and the large conversation needs to be split or moved.

References: