Resource Borrowing Cascade Depth (RBCD)

GMI-TEL-025TelecommunicationsIntegration-RequiredPattern O-6: Autonomous Coherence

Definition

Maximum number of slice-to-slice resource reallocation hops triggered by a single initial borrowing event during contention, measuring how deeply resource adjustments propagate across the slice portfolio.

The Measurement Gap

Measurement Gap and Decision

Decision Supported: Are resource borrowing events isolated (one slice adjusts, others unaffected), or do they propagate in chains that can destabilize slices far removed from the original contention?

Measurement Gap:

Resource brokers in multi-slice environments make thousands of reallocation decisions per day. Borrowing idle capacity from one slice to serve peak demand on another. Each individual decision may be locally optimal. But when Slice A borrows from Slice B, and Slice B compensates by reclaiming from Slice C, and Slice C pulls from Slice D, the cascade can destabilize slices that had no contention of their own. No current resource broker implementation tracks cascade depth: the number of sequential reallocations triggered by a single originating event. Operators see the outcome (SLA violation on Slice D) but cannot trace the causal chain back to the originating borrowing event on Slice A. Root-cause analysis of multi-slice SLA failures becomes nearly impossible without cascade visibility. RBCD measures the propagation depth of each borrowing event, tracking how many slices are affected in sequence and flagging deep cascades that represent systemic instability in the resource allocation model.

Citations:

  • 3GPP TS 28.541, "5G NRM Network Resource Model," Release 18: defines resource sharing relationships between slices but does not model cascade propagation or depth measurement.
  • ETSI GS NFV-IFA 010, "Network Functions Virtualisation: Resource Management," 2024. Describes resource pool sharing but provides no metric for cascading reallocation impact.

Formula

Plain Language: For each resource borrowing event (Slice A borrows from Slice B's idle allocation), trace whether the borrowing triggered a subsequent reallocation affecting another slice (B reclaims from C, or C reclaims from D). Count the maximum chain depth for each originating event. Track the distribution of depths.

Notation:

RBCD_event = max(hop_count) for the reallocation chain triggered by a single borrowing event

RBCD_period = P95(RBCD_event) across all borrowing events in the measurement period

Variable Definitions:

  • hop_count = number of sequential slice-to-slice resource reallocations causally linked to a single originating borrowing event (count)
  • P95 = 95th percentile of cascade depth across all events: we care about the worst cases, not the average (percentile)
  • An event with hop_count = 1 means the borrowing was absorbed with no propagation. hop_count = 3 means three slices were affected in sequence.

Units:

Integer (hop count). Lower is better.

Sign Convention:

RBCD of 1 means borrowing events do not propagate. RBCD of 4 or higher means resource adjustments cascade across multiple slices, creating systemic instability risk.

Edge Cases:

  • Parallel cascades (one borrowing event triggers simultaneous reallocations in two independent slices): count as maximum depth of either branch, not the sum.
  • Cascades that loop back to the originating slice: flag as circular. These are oscillation events and should be cross-referenced with NOOR (GMI-TEL-001).
  • Extremely rapid cascades (less than 100ms total propagation): the entire cascade may appear as a single reallocation in coarse-grained logs. Requires sub-second logging resolution.
  • Minimum sample: 50 borrowing events per measurement period.

Proposed formula, pending validation pilots.

At a Glance

Unit
Count
Trend Direction
Lower
Reporting Frequency
Monthly
Category
Inter-Slice Resource Equity and Cascade Risk
← Guaranteed Minimum Erosion Index (GMEI)Deferral Portfolio Risk Accumulation Index (DPRAI) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment