Slice SLA

Defines the minimum and peak data rates a slice must deliver. This is typically specified per UE or per slice aggregate.

• Guaranteed Bit Rate (GBR): A fixed bandwidth commitment for constant-throughput services like real-time video.
• Maximum Bit Rate (MBR): The upper limit on data rate, used for traffic shaping and preventing resource hogging.
• Aggregate Slice Throughput: The total throughput capacity available to all UEs within the slice instance simultaneously.

A violation occurs when the measured throughput drops below the GBR for a defined measurement window.

Specifies the maximum acceptable one-way or round-trip delay for a data packet traversing the slice. This is the defining metric for URLLC slices.

• User Plane Latency: The delay between the UE and the User Plane Function (UPF) anchor point.
• E2E Latency: The total delay from the application client to the application server, including transport and core network.
• Jitter: The variation in packet delay, critical for deterministic industrial control loops.

Latency budgets are often defined at the 99.999th percentile to guarantee performance for the vast majority of packets.

Quantifies the percentage of time the slice is operational and the probability of successful data delivery within the latency budget.

• Slice Availability: The uptime percentage, often expressed as 'number of nines' (e.g., 99.999% allows ~5 minutes of downtime per year).
• Packet Success Rate: The fraction of packets successfully delivered within the latency constraint. A URLLC slice may require 99.999% reliability for 32-byte packets.
• Mean Time Between Failures (MTBF): The predicted elapsed time between inherent failures of the slice during operation.

These metrics are measured end-to-end, and a failure in any sub-component (RAN, transport, core) counts against the SLA.

Defines the maximum number of concurrent users or devices the slice must support within a geographic area without performance degradation.

• UE Density: The maximum number of active UEs per square kilometer the slice can serve simultaneously.
• PDU Session Count: The total number of concurrent Protocol Data Unit sessions the slice control plane can manage.
• Connection Density: A 5G KPI requiring support for 1 million devices per km², relevant for massive IoT slices.

Exceeding these limits triggers slice admission control, where new session requests are rejected to protect existing SLA commitments.

Specifies the acceptable interruption time and handover success rate when a UE moves between cells while connected to the slice.

• Handover Interruption Time: The duration of a data path switch during a handover, targeted at 0 ms for seamless mobility in URLLC.
• Handover Success Rate: The percentage of attempted handovers that complete without a radio link failure.
• Maximum UE Speed: The highest velocity at which the slice guarantees its performance metrics (e.g., up to 500 km/h for high-speed train slices).

These metrics are critical for slices serving autonomous vehicles or high-speed rail, where a failed handover can be catastrophic.

Emerging SLA parameters that quantify the slice's power consumption relative to its delivered performance, driven by sustainability mandates.

• Energy Efficiency (EE): Measured in bits per Joule, defining the useful data transmitted per unit of energy consumed by the slice.
• Slice Carbon Footprint: The total greenhouse gas emissions attributable to the slice, calculated from its energy consumption and the grid's carbon intensity.
• Sleep Mode Ratio: The percentage of time network elements serving the slice spend in low-power states during low traffic periods.

These metrics align the tenant's operational requirements with the operator's net-zero commitments, creating a shared responsibility for energy optimization.

The gatekeeping mechanism that determines whether a new PDU session can be established within a slice without violating existing SLAs. It evaluates resource availability against the Guaranteed Bit Rate (GBR) and latency commitments defined in the contract.

• Rejects sessions that would cause SLA breaches
• Uses real-time telemetry on PRB utilization
• Prevents the 'noisy neighbor' problem in multi-tenant slices

An automation framework that continuously monitors slice KPIs against SLA targets and executes corrective actions without human intervention. A policy-driven controller compares real-time latency, throughput, and availability data to the contracted thresholds.

• Monitor-Analyze-Plan-Execute (MAPE) loop
• Triggers slice remapping or resource scaling on violation
• Essential for maintaining URLLC slice guarantees

The technical enforcement of the 'no cross-impact' clause in an SLA. Isolation ensures that a Denial-of-Service attack or traffic storm in one slice cannot degrade the performance of another slice sharing the same physical infrastructure.

• Hard isolation: Dedicated PRBs and FPGA resources
• Soft isolation: Strict quota enforcement via scheduling
• Validated through adversarial resilience testing

The dynamic scaling capability that allows a slice to meet its SLA throughput commitments during demand spikes while shrinking to conserve energy during troughs. This directly impacts the operator's cost-to-serve and the tenant's perceived performance.

• Horizontal scaling: Adding more CNF instances
• Vertical scaling: Increasing CPU/RAM for existing CNFs
• Governed by Resource Overbooking policies

The 5G core's AI-powered prediction engine that provides the predictive QoS analytics necessary for proactive SLA management. NWDAF forecasts potential SLA violations before they occur, enabling pre-emptive resource reallocation.

• Predicts slice load and user mobility patterns
• Feeds analytics to the Slice Orchestrator
• Enables predictive, not just reactive, SLA assurance

A slice type explicitly configured with dedicated, non-negotiable resources to fulfill strict SLA commitments. Unlike non-GBR slices that rely on best-effort delivery, a GBR slice reserves physical resource blocks and core network capacity.

• Suitable for industrial automation and real-time video
• SLA typically specifies a minimum guaranteed flow bit rate (GFBR)
• Admission control is mandatory for GBR slices