Getting Started — 30-Minute Tour¶

Goal: by the end of this page you will (1) understand what CSA-in-a-Box is, (2) know which deployment path matches your scenario, and (3) have a working dev environment that can run the platform locally.

If you only have 5 minutes, jump to Quickstart. If you want to deploy to Azure now, jump to Tutorial 01 — Foundation Platform.

1. What is CSA-in-a-Box?¶

CSA-in-a-Box (Cloud-Scale Analytics in a Box) is a reference implementation of an enterprise analytics + AI + data-management platform on Microsoft Azure. It bundles:

It is not a turnkey SaaS — it is opinionated open-source IaC + reference code you fork into your tenant.

2. Pick your path (decision tree)¶

flowchart TD
    Start[I want to...] --> A{Evaluate the<br/>platform?}
    A -->|Yes| Eval[Read ARCHITECTURE.md<br/>+ Use Cases section<br/>+ skim 2-3 ADRs]
    A -->|No| B{Deploy to Azure?}
    B -->|Yes| C{Commercial<br/>or Government?}
    C -->|Commercial| Comm[Tutorial 01<br/>Foundation Platform<br/>+ DLZ Bicep]
    C -->|Government| Gov[deploy/bicep/gov<br/>+ Gov Service Matrix]
    B -->|No| D{Contribute or<br/>extend?}
    D -->|Yes| Dev[CONTRIBUTING.md<br/>+ Developer Pathways<br/>+ local dev below]
    D -->|No| E{Migrate from<br/>another platform?}
    E -->|Yes| Mig[See Migrations section:<br/>AWS, GCP, Snowflake,<br/>Databricks→Fabric,<br/>Teradata, Hadoop, Palantir]
    E -->|No| Q[Open an issue<br/>or chat with the docs Copilot]

3. Local development setup (10 minutes)¶

Prerequisites¶

Clone + bootstrap¶

git clone https://github.com/fgarofalo56/csa-inabox.git
cd csa-inabox

# Python venv with all platform deps
python -m venv .venv
source .venv/bin/activate          # Windows: .venv\Scripts\activate
pip install -e ".[dev,ai,governance]"

# Verify
make typecheck                      # mypy: should report 0 errors
make test                           # pytest: ~1271 tests, all pass
mkdocs serve                        # docs at http://localhost:8000

Authenticate to Azure (only needed if deploying)¶

az login --tenant <YOUR_TENANT_ID>
az account set --subscription <YOUR_SUBSCRIPTION_ID>

# If deploying to Azure Government:
az cloud set --name AzureUSGovernment
az login --tenant <YOUR_GOV_TENANT_ID>

4. Pick your path through the docs¶

5. Common first-day questions¶

With pre-existing networking and identity (typical enterprise), a fresh DLZ deploys in ~45 minutes end-to-end. From a blank subscription with no parent ALZ, plan half a day for the first deploy because you'll iterate on parameter files. Subsequent deploys are idempotent and run in 5–10 minutes.

A dev tier (small Synapse pool, single-node Databricks Standard, B-series Functions, basic AI Search) runs roughly \(800–1,500/month** if left up 24×7. Most teams pause Synapse + Databricks outside business hours and run closer to **\)300–500/mo dev. Prod tiers scale with workload — see each example's deploy/params.prod.json for sizing assumptions.

Yes. The deploy/bicep/gov/ overlay handles MAG/USGov-specific service availability. See the Government Service Matrix for which features are GA / preview / unavailable per cloud (Public, USGov, USGov Secret, China).

See Reference Architecture — Fabric vs Synapse vs Databricks and ADR 0010 — Fabric Strategic Target. Short version: Synapse + Databricks are the production backbone today; Fabric is the strategic forward path for net-new workloads, particularly Real-Time Intelligence and Direct Lake semantic models.

Microsoft Purview is the catalog of record (see ADR 0006). Each data product owns its YAML contract under examples/<vertical>/contracts/ and a sync job pushes contracts → Purview classifications + lineage. See Best Practices — Data Governance and ADR 0012 — Data Mesh Federation.

Out of scope — see ADR 0011 — Multi-Cloud Scope for the rationale. We document virtualization patterns (query AWS/GCP from Azure) but do not maintain parallel IaC for other clouds.

The repo is actively maintained against Azure GA features. Preview-only services are documented but gated behind feature flags. ADRs are revisited annually. The Platform Research Report is a snapshot of strategic direction.

6. Next steps¶

7. Getting help¶

• Open an issue: https://github.com/fgarofalo56/csa-inabox/issues
• GitHub Discussions: https://github.com/fgarofalo56/csa-inabox/discussions (Q&A + feature requests)
• Docs Copilot: in-page chat widget on every docs page (powered by Azure OpenAI in our DLZ — see ADR 0022)
• Security issues: see SECURITY.md (private disclosure)

See also:

• ← Previous: Documentation home
• → Next: Quickstart
• ⌂ Index: Documentation home