Nokia QoS key terms

A set of mechanisms that classify, mark, queue, and schedule traffic so that latency, jitter, loss, and bandwidth targets are met for selected flows during congestion.

A scalable QoS model that groups traffic into behavior aggregates by marking packets and applying per-hop behaviors at each node, rather than reserving resources per flow.

A per-flow QoS model that uses explicit signaling and admission control to reserve resources along a path, typically associated with RSVP.

A six-bit field in the IP header used to select the per-hop behavior a packet receives; routers read it to choose queue, drop precedence, and scheduling treatment.

The forwarding treatment a node applies to a group of packets sharing a DSCP value, defining scheduling, buffer, and drop behavior at that hop.

A PHB providing low loss, low latency, low jitter, and assured bandwidth, typically used for voice; it is implemented with strict priority scheduling and rate policing.

A PHB group with four classes and three drop precedences per class, allowing bandwidth assurance relative to other traffic while permitting preferential dropping within a class.

Eight DSCP codepoints (CS0 to CS7) defined for backward compatibility with IP precedence, commonly used as a simple eight-class marking scheme.

The original three-bit priority field in the IPv4 type-of-service byte, still used for backward-compatible classification and mapping into modern class schemes.

The point in the network where QoS markings from attached devices are accepted or overwritten, ensuring that untrusted sources cannot claim high-priority treatment.

The process of identifying packets and assigning them to a forwarding class or QoS policy entry using match criteria such as DSCP, IP address, port, or protocol.

Setting or rewriting a field such as DSCP, IP precedence, or Ethernet priority so downstream nodes can apply consistent treatment without repeating deep classification.

Metering traffic against a committed rate and taking action on excess, typically dropping or remarking out-of-profile packets; it requires limited buffering and does not delay excess traffic.

Buffering and delaying excess traffic to smooth it to a configured rate rather than dropping it, trading increased latency and buffer use for lower loss.

Measuring the arrival rate of a traffic stream, often with a token bucket, to determine whether packets are in or out of profile for policing or marking decisions.

A rate-measurement model where tokens accumulate at a committed rate up to a burst size; each packet consumes tokens, and insufficient tokens indicate out-of-profile traffic.

The guaranteed average rate a service or policer commits to carry, usually paired with a burst allowance that determines how much traffic may exceed the rate briefly.

An internal grouping in Nokia routers that determines the queue and scheduling treatment a packet receives; classification maps traffic into a forwarding class.

A Nokia router construct that maps forwarding classes into a smaller set of per-hop behaviors used on egress network ports, decoupling local classes from wire markings.

A Nokia construct that associates a set of queues with a forwarding class or interface so that identical scheduling treatment can be reused across many ports.

A QoS component that allocates bandwidth and priority among queues, defining rate, weight, and priority parameters applied to a queue, port, or aggregate.

A scheduling discipline that always serves higher-priority queues before lower ones, giving minimal delay to priority traffic but risking starvation of lower classes.

A scheduling method that shares bandwidth among queues in proportion to configured weights, providing fairness and predictable relative capacity during congestion.

An active queue management algorithm that drops packets probabilistically before a buffer fills, with drop likelihood influenced by DSCP or drop precedence to signal congestion early.

A simple queue overflow behavior that discards all arriving packets once the buffer is full, often causing global synchronization and unfair loss across flows.

A marking attribute that orders packets within the same class so that lower-precedence packets are discarded first when congestion builds, preserving higher-value traffic.

A priority queue configuration that gives delay-sensitive traffic preferential scheduling while policing it to a configured rate to prevent starvation of other queues.

A named Nokia configuration object containing classification, marking, policing, and queue mapping rules that is applied to an interface or service to enforce treatment.

A Nokia policy defining how DSCP or dot1p values are interpreted on ingress and which values are written on egress, controlling marking behavior at the trust boundary.

The point on a Nokia service where customer traffic enters or leaves, and where ingress and egress QoS policies are typically attached for per-service enforcement.

Applying per-service policies so that each customer or service receives the contracted treatment, including rate limits, marking, and scheduling across shared infrastructure.

A contract specifying measurable performance commitments such as availability, latency, jitter, loss, and bandwidth, against which delivered QoS is monitored and reported.

A decision process that accepts or rejects a new flow or service based on whether sufficient resources remain to honor existing commitments.

A scalar value in 5G that references a standardized set of QoS characteristics such as resource type, priority, delay budget, and error rate for a QoS flow.

The finest granularity of QoS treatment in 5G, identified by a 5QI and optionally an allocation and retention priority, carried over radio and transport segments.

A scalar value in LTE that identifies standardized bearer characteristics including resource type, priority, packet delay budget, and packet error loss rate.

The bit rate guaranteed to a 5G QoS flow by the network, used to ensure that guaranteed-bit-rate flows receive at least their committed capacity.

Variation in packet inter-arrival delay over a path; excessive jitter degrades real-time voice and video and is controlled by queuing, shaping, and de-jitter buffering.

The time a packet takes to travel from source to destination, including serialization, propagation, queuing, and forwarding delays; queuing delay is the part QoS most directly controls.

Performance monitoring techniques that use timestamped or colored probe packets to measure delay and loss per class on live traffic without injecting separate test streams.