{{ summaryAnnouncement }}
{{ summaryHeading }}
{{ summaryPrimary }}
{{ summaryLine }}
{{ badge.label }}{{ badge.value }}
{{ formatNumber(model ? model.total_reserved_percent : 0, 0) }}% P{{ row.priority }} {{ formatNumber(row.share, 1) }}%
QoS bandwidth allocation controls
QoS parent rate and traffic class allocations
Class percentages are applied to this policy rate after optional shaping headroom.

Traffic classes:

Add every explicitly reserved class. The calculator reports any remaining parent share as best effort.
ClassShareBehaviorPriorityNoteRemove
%
{{ workflowFeedback }}
Zero is neutral.
%
One is neutral and keeps burst ceilings equal to reserved rates.
×
Zero disables this optional review and is neutral.
%
{{ ledgerExportAnnouncement }}
ClassBehaviorShareReservedBurst ceilingPriorityNoteCopy
{{ row.name }}{{ row.behavior }}{{ row.share }}{{ row.reserved }}{{ row.burst }}{{ row.priority }}{{ row.note }}

{{ check.title }}

{{ check.status }}

{{ check.reading }}

{{ check.action }}

{{ checksAnnouncement }}
{{ chartExportAnnouncement }}

Quality of Service (QoS) planning decides how a congested egress link should share scarce capacity. The useful rate is the policy's real parent rate, often a shaper below the physical interface speed. Class percentages are applied to that usable rate, so an incorrect parent value makes every reserved bandwidth figure wrong by the same proportion.

A reservation is not a promise that an application will always receive that rate. Classification must put packets into the intended class, the scheduler must implement the policy as expected, and ingress policing or admission control may be needed to keep high-priority traffic within its engineered limit. When the link is not congested, many schedulers allow classes to use capacity beyond their minimum reservation.

Reserved share
The percentage of usable parent bandwidth assigned to a named class.
Best-effort headroom
The unreserved share left for default or unmatched traffic when named reservations total less than 100%.
Priority pressure
The combined share marked for strict or priority treatment, compared with an operator-selected review threshold.
Burst ceiling
A planning multiple of the reserved rate. It is not a token-bucket calculation or a device configuration by itself.

The class behavior label and priority number describe policy intent, but this percentage model does not simulate a vendor scheduler. Priority order does not change the arithmetic reservation. Behavior affects the separate priority-pressure check only when a class is marked strict or priority.

Leaving some unreserved capacity can protect routing, management, default traffic, and estimation error. Reserving exactly 100% is valid arithmetic, yet it leaves no best-effort share. Reserving more than 100% creates a shortfall and should be corrected before a policy is deployed.

A bandwidth plan is a design check, not live validation. Packet counters, queue drops, latency, jitter, class matches, and the configured scheduler must still be observed on the actual device under representative load.

How to Use This Tool:

Model the egress policy from its parent rate downward, using one row for each explicitly reserved class.

  1. Enter Parent bandwidth and select Kbps, Mbps, or Gbps. Use the shaped or contracted rate that actually feeds the scheduler.
  2. Add each traffic class with a unique name, percentage share, behavior, priority from 1 through 8, and an optional note. Up to 12 classes are accepted.
  3. Set Shaping headroom when the usable policy rate should stay below the entered parent rate. The allowed range is 0% through 50%.
  4. Use Burst ceiling multiplier only as a planning comparison from 1 through 10. Set Priority review threshold above zero when strict and priority reservations should trigger a policy review.
  5. Correct any class warning before interpreting the ledger. Duplicate names, unsupported behaviors, priorities outside 1 through 8, and shares outside 0% through 100% are rejected.
  6. Check Total reserved, Best effort, Shortfall, and Strict and priority share. A ready result still needs device-specific configuration and traffic testing.

Interpreting Results:

Usable parent is the bandwidth left after optional shaping headroom. Each class reservation and the automatic best-effort remainder come from that rate. Compare Mbps as well as percentages, because the same policy percentages produce very different capacity on 100 Mbps and 1 Gbps links.

Overallocated means named shares total more than 100%; the displayed shortfall is the amount that cannot fit. Fully allocated means the total is exactly 100% within floating-point tolerance and best effort is zero. A priority review appears only when the threshold is enabled and the combined strict-plus-priority share is strictly greater than it. Equality does not trigger the review.

The burst values are arithmetic ceilings from the selected multiplier. Confirm burst units, policer or shaper semantics, queue limits, and borrowing behavior in the target platform before translating them into configuration.

Technical Details:

Differentiated Services supplies packet classification and per-hop treatment building blocks, while an operator's scheduler and policy determine actual bandwidth behavior. This model is deliberately narrower: it converts class shares into rates, reports unused or oversubscribed capacity, and highlights the amount assigned to strict and priority behavior.

Formula Core

Parent bandwidth is normalized to Mbps. Shaping headroom reduces the rate before any class share is applied.

U=P(1h100)

Each reserved rate is its percentage of usable parent bandwidth. The burst ceiling multiplies that reserved rate without changing the reservation.

Ri=Usi100 Ci=mRi

The remainder and shortfall are complementary. Only one can be positive for a valid calculation.

E=max(0,100isi) Q=max(0,isi100)
QoS allocation formula symbols
SymbolMeaningUnit
PNormalized parent bandwidthMbps
hShaping headroompercent
UUsable parent bandwidthMbps
siShare assigned to class ipercent
RiReserved bandwidth for class iMbps
m, CiBurst multiplier and resulting ceilingmultiplier, Mbps
E, QBest-effort remainder and over-allocation shortfallpercent

Rule Core

QoS allocation status rules in evaluation order
StatusExact conditionMeaning
OverallocatedTotal named share > 100%Reservations exceed usable parent capacity
Priority reviewThreshold > 0 and strict-plus-priority share > thresholdPriority pressure exceeds the operator's review line
Fully allocatedAbsolute remainder < 0.0000000001%No best-effort share remains
ReadyNone of the earlier conditions appliesThe arithmetic plan fits with some remainder

Over-allocation takes precedence over the priority review. Priority values from 1 through 8 do not change the bandwidth formula, and the behavior category changes arithmetic only by deciding whether a class contributes to the strict-plus-priority review sum.

Worked Examples:

Shaped 100 Mbps policy

A 0.1 Gbps parent with 5% shaping headroom leaves 95 Mbps usable. Shares of 10% voice, 20% video, and 50% data reserve 9.5, 19, and 47.5 Mbps. The classes total 80%, leaving 19 Mbps or 20% for best effort. With a 1.5 burst multiplier their arithmetic ceilings become 14.25, 28.5, and 71.25 Mbps. A 40% priority threshold does not trigger because strict-plus-priority share is 30%.

References: