P9 is a Windows-native network simulation system for shaping latency, loss, jitter, bandwidth, reordering, and duplication. Its first-party Packet Warp dataplane connects controlled conditions to deterministic evidence.
The interface keeps condition controls, operator state, and network telemetry in one governed chamber. The capture below is a real local P9 application frame, not a generated product mockup.
01Condition controls
02Operator state
03Network telemetry
Authentic application capture. The shown state is disarmed, offline, and has no profile loaded. It is not live telemetry and does not demonstrate signed-driver qualification.
Operating purpose
Turn network variability into a controlled experiment.
Define the condition that matters, observe nominal behavior, apply a reusable profile, replay the same boundary, compare the outcome, and retain the evidence. P9 keeps operator intent, execution, observation, and recovery context connected through that cycle.
Condition definedSet the impairment boundary.
Baseline observedObserve nominal behavior.
Profile appliedActivate reusable policy.
Condition replayedRepeat the same condition.
Outcomes comparedInspect outcome differences.
Evidence preservedRetain traceable evidence.
This sequence describes the operating model. It is not a benchmark result or a claim that every environment will produce the same outcome.
System architecture
One governed path from intent to verdict.
This architecture model describes the implemented control and dataplane boundary. Its motion is explanatory, not a live traffic visualization.
01 / Intent
Operator application
Profiles, condition policy, activation state, and visible controls.
02 / Governance
PARADIGM control + evidence plane
Deterministic state, bounded memory, replay records, and operator explanations.
03 / Contract
Packet Warp protocol v5
Versioned configuration, capability negotiation, counters, and verdict exchange.
04 / Dataplane
WFP classify + bounded verdict worker
Transactional lifecycle, deadline-aware handling, and explicit fail-open behavior.
05 / Path
Windows network path
Traffic outcome remains subject to live-host, adapter, and signing qualification.
This summary describes the reviewed control model. It does not define a public API, hosting topology, entitlement rule, or deployment guarantee.
Evidence ledger
Measured claims, labeled by scope.
Every figure below is tied to an existing repository record. Control-plane and synthetic results are not presented as packet, NIC, or wire-rate evidence.
Native / dataplane gates
400
Focused tests passed
Recorded against the Packet-Warp-only MSVC build, including build, parity, awareness, HyperPath, and protocol gates.
Deterministic replay
7 / 7
Control-plane scenarios passed
Nominal, degraded, critical, recovery, fault, and insufficient-evidence behavior with repeated digest checks.
Synthetic matrix
1.282x
Median clean-binary speedup
Across 54 policy, size, batch, and worker cases: 46 of 54 faster; 8 at or below the legacy result.
Scope boundary
The 54-case performance matrix is synthetic. It is not NIC or wire-rate evidence, and it does not establish universal performance. The seven-scenario replay and recorded latency measurements are control-plane evidence, not Packet Warp dataplane throughput.
Core capabilities
One command continuum. Four operational systems.
P9 connects packet conditioning, repeatable profiles, operator context, and control integrity in one Windows-native system.
Condition engine
Shape packet behavior.
Build controlled conditions for the network variables that reveal fragile system behavior before deployment.
P9 is framed for Windows deployment. Exact supported Windows editions and builds, processor, memory, storage, network-adapter, privilege, signing, and deployment requirements have not been approved for publication.
Operating system
Exact supported Windows editions and builds pending technical review.
Compute and storage
Processor, memory, and storage requirements pending technical review.
Network and privileges
Adapter, privilege, and policy requirements pending technical review.
Deployment and signing
Environment qualification and signing requirements remain release gates.
System requirements pending technical review
Engineering readiness
Maturity states, kept separate.
Qualified user-mode evidence, production-signing status, and theoretical research are not interchangeable claims.
WorkstreamEngineering stateRelease gate
Azure-signed user modeClean-host qualifiedEvidence available
Packet Warp DriverEngineering completeMicrosoft production signing pending
Null Point HyperPathArchitecture researchTheoretical
Microsoft has not endorsed or production-signed Packet Warp. Null Point HyperPath is a research direction, not released capability.
Release boundary
What remains before a production-driver claim.
Code inspection and user-mode tests do not substitute for a signed disposable-host run. Packet Warp remains a development asset until the external qualification sequence is complete.
Pending 01Compliant signing path
Complete the certificate, organization, and Microsoft Hardware Developer Program prerequisites.
Pending 02Disposable-host execution
Load only on an approved lab target and collect real WFP traffic evidence.
Pending 03Verifier and platform qualification
Exercise Driver Verifier, applicable HLK or WHCP gates, adapter churn, sleep/resume, overload, and rollback.
Pending 04Signed release evidence
Require returned signing artifacts, reproducible packaging, installation, upgrade, uninstall, and rollback records.
Technical evaluation
Evaluate P9 with technical context.
Evaluation begins with the operating need, Windows deployment context, intended scale, and validation goals. Qualification evidence and release readiness are reviewed before an evaluation path is planned; access, pricing, and deployment remain conditional.
Define the operating need · Review qualification evidence · Plan the evaluation path
