System Context (L1)

This view explains where IMP sits in your platform and what it depends on.

Purpose

IMP gives Kubernetes-native control over Firecracker microVM workloads. It keeps VM lifecycle and networking in Kubernetes APIs, while executing runtime operations on worker nodes.

Primary Actors

• Platform Admin: installs and configures CRDs, controller, and policies.
• Cluster User: creates ImpVM and related resources to request workloads.
• IMP Control Plane: operator reconciles desired state and status.
• IMP Data Plane: agent performs node-level runtime and network actions.

External Dependencies

• Kubernetes API + etcd: source of truth for desired and observed state.
• CNI stack (Cilium-first): network policy, IPAM, observability.
• OCI image sources and storage backends used by workload artifacts.
• Prometheus-compatible stack for metrics and alerting.

Boundaries

Inside IMP boundary:

• CRD reconciliation and orchestration logic.
• VM runtime and lifecycle automation.
• Networking setup and status reporting.

Outside IMP boundary:

• Tenant application logic running inside guest VMs.
• Cluster-level security posture and baseline hardening.
• Infrastructure lifecycle of underlying nodes.

Success Criteria

• Declarative VM workflows from Kubernetes resources.
• Predictable reconcile behavior (idempotent, retry-safe).
• Clear status, events, and metrics for operations teams.