Federated Query Platform for ML at Scale: Architecture and Multi-Tenancy

Back to Devoured - June 18, 2026

DEVOURED

Federated Query Platform for ML at Scale: Architecture and Multi-Tenancy

Data infrastructurecloud Medium

Guidewire deployed Apache Trino to create a federated query layer across Iceberg, Redshift, and S3, exposing gaps in S3 governance.

What: Guidewire implemented Trino for federated SQL querying across diverse storage backends (Iceberg, Redshift, OpenSearch). While ABAC (Attribute-Based Access Control) worked well, the team found that S3-level security lacked granular governance.

Why it matters: Engineering organizations attempting to centralize access to heterogeneous data stores often find that standard RBAC/ABAC tools fail to provide sufficient control at the object-storage level.

Decoder

Federated Query: A technique that allows users to run a single query across multiple independent data sources without moving the data into a central repository.
Iceberg: An open table format for large analytic datasets that supports schema evolution and partition evolution.
RBAC: Role-Based Access Control, a method of restricting system access to authorized users based on their assigned roles.

Original article

Guidewire used Apache Trino to federate SQL across Iceberg, Redshift, OpenSearch, and customer S3 buckets, speeding exploratory ML workflows. Domain catalogs and Lake Formation ABAC plus Trino RBAC handled isolation, but Trino's S3 layer exposed a governance gap.

Explore all 55 articles in today's digest