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

Engineering notes

Documentation & tests

This page exists because documentation is part of the deliverable, not an afterthought. It states plainly what is simulated, what production would require, and how the logic behind these screens is verified.

Architecture
Front-end only, deliberately.

The application is a TypeScript React front end built with TanStack Router, Tailwind and shadcn/ui components. All demonstration data lives in one typed module with pure helper functions, so the derived metrics on every page come from testable logic rather than hard-coded strings.

There is intentionally no backend, no database driver and no network call. A portfolio piece should not ship a database connection it cannot secure, and claiming a live Oracle integration that does not exist would be dishonest. The screens model the operational surfaces I build and run in real engagements.

Demo vs production boundary
Read this before interpreting any number on the site.

Data

Demo

Hand-authored synthetic records in a single typed module.

Production

Oracle schemas fed by change capture from source systems.

Execution

Demo

No jobs run; run histories are static fixtures.

Production

Scheduler-driven jobs with checkpointing, retries and alerting.

Metrics

Demo

Fixed plan text, wait profiles and timings written for illustration.

Production

Collected from the database's own performance views and a metrics store.

Access

Demo

Fully public, read-only pages. No authentication.

Production

Authenticated access, least-privilege roles, audited privileged actions.

Secrets

Demo

None exist. Connection details are placeholder strings.

Production

Vault-issued short-lived credentials, rotated automatically.

Test suite
Unit tests in src/lib/lab-data.test.ts, run with vitest run.
  • successRate / rejectRateQuality maths, including divide-by-zero guards.
  • isWithinSlaSLA boundary behaviour at exactly the threshold.
  • speedupTuning gain multiplier for each case study.
  • fleetHealthFleet rollup consistency against the catalogue.
  • formatCompactK/M formatting thresholds.
  • catalogue integrityUnique slugs, classification on every column, no credential-like strings.
Conventions
  • Semantic design tokens only; no ad-hoc colour values in components.
  • Derived numbers come from pure functions, and pure functions are tested.
  • Every page states its simulated nature where a reader could misread it.
  • Accessible structure: one H1 per page, labelled navigation, real table semantics.
  • Each route carries its own title and description metadata.
  • Lint, tests and a production build all run clean before delivery.