Stage 4 of 6

Unified Data Foundation

The single source of truth: every event stored once, in an open format, with metadata and governance applied consistently across raw, curated, and analytics-ready data.

🗃 CXOS Data Lakehouse

Apache Iceberg + S3 Compatible Storage

How to Consume

Cxos.Foundation.Infrastructure · Both

-- Any Iceberg-compatible engine attaches directly, no CXOS API in the read path:
SELECT * FROM curated.identity_stitched_events
WHERE event_date >= '2026-07-26';
-- location: abfss://lakehouse@cxosdata.dfs.core.windows.net/curated/events

🧠 Metadata Layer

Central Nervous System

How to Consume

Cxos.Foundation.Api · Query

GET /v1/catalog/tables/curated.identity_stitched_events
{ "owner_team": "platform-engineering", "row_count_estimate": 2847193021,
  "lineage": ["raw.events"], "last_updated": "2026-08-02T09:00:00Z" }

🔒 Governance & Security

How to Consume

Cxos.Foundation.Api · Both

// Policy is checked automatically on every read via the shared policy library —
// not something a caller invokes separately. To inspect a decision:
GET /v1/governance/access-decision?principal=agt_302&resource=curated.customer_profile
{ "decision": "allow", "constraints": { "column_mask": ["email", "phone"] } }

🔌 Platform Connectors

The engine attachments every Iceberg-compatible consumer connects through — no data copy required.

Connector PackageConsumer
Iceberg REST Catalog Query & Analytics Engine (DataFusion) · ad hoc Spark/Trino
Snowflake external tables Snowflake / BigQuery (Destinations & Activation)
Databricks Iceberg connector Databricks / Redshift (Destinations & Activation) · Predictive Models
Azure Purview scanners Catalog · Lineage · Data Dictionary
Real World Example

All events are stored in the data lakehouse once, in open formats. You can query raw events from yesterday or aggregated data from last year without moving or duplicating data.