Simulated environment. All schemas, metrics, and run histories are synthetic; no live Oracle instance, credentials, or customer data are used.

Oracle Secure Data Engineering Lab

Architecture

Reference architecture and the demo boundary

The layered design below is what I would build on a real Oracle estate. This application demonstrates the design, the metrics it would emit and the operating practice around it — it does not connect to Oracle, ODI, Informatica or GoldenGate.

End-to-end flow

Concept diagram
  ┌──────────────┐   ┌───────────────┐   ┌────────────────┐   ┌──────────────┐
  │ Source feeds │──▶│ Landing (STG) │──▶│ Conform (PL/SQL)│──▶│ Serve (facts)│
  └──────────────┘   └───────────────┘   └────────────────┘   └──────┬───────┘
        watermark          DQ gate            SCD2 + MERGE           │
                                                                     ▼
                                            ┌─────────────┐   ┌──────────────┐
                                            │ Audit sink  │◀──│  Replication │
                                            └─────────────┘   └──────────────┘
                                              write-once        CDC / standby

Layer by layer

Ingest & landing
AR-01
File/API feeds · Landing tables · Watermark checkpoints

Demonstrated here

Simulated feed metadata, arrival-window checks and checkpoint semantics.

Production shape

External tables or an ODI/Informatica load plan writing to a landing schema with idempotent watermarks.

Transform & conform
AR-02
PL/SQL packages · Set-based MERGE · SCD2 dimensions

Demonstrated here

Annotated PL/SQL and MERGE patterns rendered as documentation, not executed.

Production shape

Versioned PL/SQL packages deployed through CI, orchestrated by ODI mappings or an equivalent scheduler.

Serve & analytics
AR-03
Partitioned facts · Materialised aggregates · Read-only reporting replica

Demonstrated here

Partitioning strategy, aggregate design and replica lag reporting.

Production shape

Range-partitioned facts with local indexes, refreshed aggregates and a reporting standby.

Replication & resilience
AR-04
Log-based CDC · Physical standby · Backup & restore rehearsal

Demonstrated here

Lag budgets, drill outcomes and RPO/RTO targets against synthetic runs.

Production shape

GoldenGate-style CDC plus Data Guard-style standby, with scheduled restore validation.

Governance & audit
AR-05
Column classification · Unified audit sink · Retention proofs

Demonstrated here

Control catalogue, classification labels and a synthetic audit evidence trail.

Production shape

Unified auditing to a write-once sink, quarterly recertification and automated retention proofs.

How this application itself is built

InterfaceReact 19 · TypeScript · Tailwind v4 · shadcn/ui
RuntimeTanStack Start (SSR) on an edge worker
Data layerTyped synthetic fixtures, no network calls
Quality gateESLint · Prettier · Vitest unit tests · production build

Implementation status: all pages are complete and driven by typed synthetic fixtures. Deterministic quality, reconciliation, replication and backup helpers are unit tested. Live database connectivity, a real audit sink and scheduled orchestration are documented as architecture concepts and are deliberately out of scope for this portfolio lab.