Phased Admission Trial¶
Protocol — instantiates Displacement-Aware Capacity Admission
A staged rollout of the entrant with measurement gates, rollback authority, and incumbent impact review.
When the displacement forecast is uncertain, you don't have to bet the whole substrate on it — you can admit the entrant a slice at a time and let each slice earn the next. Phased Admission Trial is the protocol that does this: it breaks the entrant's claim into staged increments, admits the first small increment, measures the actual displacement against the forecast at a gate, and only advances to the next increment if the gate passes. Its defining move is incremental, evidence-earned admission — the entrant's full claim is a hypothesis, and each stage buys real data on it before more substrate is committed. It is the graduated on-ramp: it decides how much to admit next based on what the last increment did, converting a single high-stakes yes/no into a sequence of small, reversible, checkable steps.
Example¶
A platform team must let a new event-ingestion service onto a shared Kubernetes cluster that already runs the company's checkout and search workloads. The service's owners claim it needs modest resources and won't disturb neighbors, but shared-cluster noisy-neighbor effects are notoriously hard to predict. So the team runs a Phased Admission Trial. The entrant's claim — steady-state CPU, memory, and network budget — is decomposed into four stages: 5% of production traffic, then 25%, 50%, and full. At 5% the service runs as a canary alongside the incumbents, and a gate is defined in advance: checkout p99 latency must stay under its SLA, search error rate must not rise, and the cluster's scheduling-pressure metric must stay in the green band.[n1]
At the 25% stage the gate catches something the forecast missed: the ingestion service's bursty write pattern is evicting cache pages that search depends on, and search latency drifts toward its limit. Because admission was staged, only a quarter of traffic is affected, and rollback authority is already in the protocol — the team halts the advance, holds at 25%, and sends the service back to tune its batching before earning the next increment. The incumbents were never put fully at risk, because the entrant had to prove each step before taking the next.
How it works¶
The protocol's distinguishing logic is the gated increment: measure the real thing at small scale before scaling it up:
- Decompose the claim into stages. Break the entrant's full request into increments sized so a failure at any stage is survivable for the incumbents.
- Define pass/fail gates on incumbent impact. Before each stage, fix the incumbent-facing metrics and thresholds that must hold to advance — the gate tests the displacement, not just the entrant's own health.
- Run the monitoring loop each stage. Instrument incumbent and entrant metrics continuously through the increment, comparing observed displacement to the assessment's forecast.
- Advance, hold, or roll back on evidence. The admission rule for the next increment is conditional on this one's gate; a failed gate holds or reverts rather than proceeding.
Tuning parameters¶
- Stage granularity — how many increments and how large each. Fine stages catch trouble early and cheaply but lengthen the rollout; coarse stages are fast but bet more per step.
- Gate strictness — how tight the pass thresholds are. Strict gates protect incumbents but may stall a benign entrant on noise; loose gates advance smoothly but let real displacement through.
- Dwell time — how long each stage runs before the gate is read. Longer dwell captures periodic peaks and slow effects; shorter dwell speeds rollout but can miss a weekly spike.
- Rollback scope — whether a failed gate holds at the current stage or reverts to the previous one. Reverting is safer; holding preserves progress but leaves the current displacement in place.
When it helps, and when it misleads¶
Its strength is that it makes admission reversible and cheap to be wrong about — the entrant proves itself against the incumbents at small scale, so a bad forecast costs a fraction of the substrate instead of all of it. It naturally supports the archetype's core caution: benefits are measured, displacement is diffuse, and staging forces the diffuse harm to show itself before the commitment is total.
Its failure mode is the trial whose gates watch the entrant's health but not the incumbents' — a canary that reports "the new service is fine" while checkout quietly slows, because the gate never measured checkout. Ratchet pressure is the other trap: once an entrant is at 50% and integrated, holding or reverting gets politically harder at each stage, so a nominal gate becomes a rubber stamp. The guarding discipline is to gate on incumbent-facing displacement metrics, pre-authorize the hold/rollback so advancing is not the default, and keep dwell long enough to see periodic strain. The trial watches and advances; it does not define the pre-committed stop-and-reverse thresholds for an already-fully-admitted entrant — that is Moratorium and Reversal Gate.
How it implements the components¶
The trial is a staged-rollout protocol filling three components:
entrant_or_expansion_claim— decomposes the newcomer's full request into survivable increments, the unit each stage admits.phase_in_monitoring_loop— instruments incumbent and entrant metrics through each increment, comparing observed to forecast displacement.admission_or_expansion_rule— the gate that makes each next increment conditional on the prior one's measured impact.
It does not maintain the standing, always-on displacement_accountability_record that persists after the rollout ends — that is Crowding-Out Monitoring Dashboard. It does not set the protected_function_floor reserved before any admission — that is Capacity Reservation Rule.
Related¶
- Instantiates: Displacement-Aware Capacity Admission — the trial supplies the staged, evidence-gated on-ramp that admits an uncertain entrant without betting the whole substrate.
- Consumes: Crowding-Out Monitoring Dashboard supplies the per-stage displacement measurements; Moratorium and Reversal Gate supplies the rollback authority a failed gate invokes.
- Sibling mechanisms: Incumbent Use Register · Substrate Capacity Budget · Displacement Impact Assessment · Shadow Displacement Accounting · Capacity Reservation Rule · Moratorium and Reversal Gate · Offset or Relocation Plan · Crowding-Out Monitoring Dashboard
Editorial Notes¶
Form Classification¶
Form family: Experiment, Test & Rehearsal
Rationale: Phased Admission Trial operates as an active test, trial, simulation, drill, or rehearsal that generates evidence through a deliberate attempt or perturbation because it a staged rollout of the entrant with measurement gates, rollback authority, and incumbent impact review.
Independent corroboration: The frozen evidence defines Phased Admission Trial as 'A staged rollout of the entrant with measurement gates, rollback authority, and incumbent impact review', so its operative form is Experiment, Test & Rehearsal.
Nearest alternative: Protocol, Workflow & Routine — Phased Admission Trial includes features of a repeatable ordered procedure or handoff sequence that coordinates action, but its defining operation is an active test, trial, simulation, drill, or rehearsal that generates evidence through a deliberate attempt or perturbation.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Biology & Ecology
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Staged introduction with incumbent-impact monitoring is rooted in invasion ecology and biological containment.
Related originating lineages:
- Agricultural Science & Agronomy — Biosecurity and quarantine practice materially developed controlled admission of novel organisms.
- Computer Science & Software Engineering — Phased Admission Trial is rooted in computer science and software engineering: Canary-release practice stages a new entrant and gates expansion on both entrant and incumbent health.
- Engineering & Design — Engineering and design materially shaped Phased Admission Trial through reliability, physical systems, safety, and mistake-proof design.
- Environmental Science & Climate Studies — Environmental risk assessment supplied rollback gates and ecosystem-impact review.
Review resolution: Light authoritative-source research resolves the primary-origin disagreement in favor of biology and ecology. FAO: Import Risk Analysis and Ecological Invasion Pathways directly documents the defining practice or theory described in the selected origin rationale. Other listed domains are retained only where the blind reviews identify material co-development or translation; broader adoption remains separate as domain_reach=multi_domain.
Attribution caveat: The boundary with computer science and software engineering is real because that field materially developed or translated the practice, but the cited provenance places the defining form in biology and ecology.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
Notes¶
[n1] A canary release rolls a change out to a small fraction of traffic first, watches health signals, and promotes it only if they hold — named for the canary in a coal mine. The trial extends the idea from the entrant's own health to the incumbents it shares a substrate with. ↩