Data mesh & governance
Federate the data. Federate the governance.
A central data lake doesn't scale to fifty domains and twelve jurisdictions. Indentia federates: each domain owns its data products under a shared ontology, governance rules are written once and enforced everywhere, and queries cross domains without copies. One coherent picture; no central bottleneck.
Capabilities
Six pieces. One coherent governance model.
Data mesh — per-domain governance
Each business domain owns its data products: schema, contracts, quality, ownership. The platform federates over them rather than centralising. Governance becomes a property of each domain, not a bottleneck queue at a central team.
Ontology-driven domain contracts
Domains expose their data through semantic contracts grounded in a shared ontology. A "customer" in CRM and a "customer" in billing reconcile through ontology mappings — not through brittle string-matching. New domains plug in without renegotiating the whole graph.
Federated SPARQL
Query across domains without copying their data. SPARQL federation joins a graph in domain A with a graph in domain B at query time; each domain enforces its own ACLs on the pieces it owns.
Label propagation through lineage
A sensitivity label set on a source dataset travels along its lineage — through dbt models, derived datasets and Superset dashboards. Downgrade is explicit and auditable; nothing leaks silently downstream.
Policy-as-code (OPA + ODRL)
Access decisions are expressed in Rego (OPA) for runtime evaluation, and in ODRL for declarative usage policies on data products. The same policy text is auditable, version-controlled and machine-evaluated — no "we will write a process around it" gap.
Federated governance, automated
Cross-domain rules — retention, classification, jurisdiction — are written once and enforced at every domain by the same policy engine. The central team sets the frame; each domain runs within it.
Where it lives in the platform
Federation is a property of the stack, not a separate product.
Data catalogue & lineage
DataHub indexes datasets, dashboards and pipelines, with OpenLineage events from every transformation.
→ detailsFederated SQL
Trino federates over warehouses, databases, lakehouses and streams under one identity.
→ detailsEvent mesh
NATS + Pulsar carry policy decisions, lineage events and domain notifications across the federation.
→ detailsIdentity model
Zitadel-backed OIDC + ABAC over the ontology, with per-domain organisations and roles.
Written once. Enforced everywhere.
Policy that runs, not policy in PDF.
Retention, classification, jurisdictional residency, releasability — captured as code in OPA/Rego or as ODRL usage policies on each data product. Every cross-domain access decision is evaluated against the same rules, logged with the same identity, audited in the same trail. No more "the policy says X but the system does Y".