Noémi Ványi: We Skipped the OLAP Stack and Built Our Data Warehouse in Vanilla Postgres

Why It Matters

By unifying disparate data sources in Postgres, Xata gains real‑time product insights without sacrificing production performance. The low‑overhead solution accelerates data‑driven decisions for a fast‑growing SaaS.

Key Takeaways

• Centralized four data sources into single Postgres warehouse
• Used materialized views and pg_cron for fast analytics
• Database branches enable safe schema iteration
• Postgres handles analytics at small‑medium scale, deferring OLAP
• Automated nightly pipelines keep data fresh without heavy ops

Pulse Analysis

Enterprises and startups alike are reevaluating the need for heavyweight OLAP platforms when a well‑tuned PostgreSQL instance can serve as a unified analytics hub. PostgreSQL’s mature indexing, robust extensions, and native JSONB support make it a viable alternative for small‑to‑medium data volumes, especially when operational overhead must stay low. Xata’s decision reflects a broader shift toward leveraging existing transactional databases to answer product‑level questions—such as trial conversion or branch creation rates—without introducing a separate data warehouse stack. This consolidation reduces latency, cuts licensing costs, and aligns analytics directly with the core application stack.

The Xata implementation hinges on three PostgreSQL features: materialized views, pg_cron, and database branching. Materialized views flatten event payloads from PostHog, turning nested JSONB into columnar formats that query engines can scan efficiently. pg_cron automates daily refreshes, ensuring the analytical layer stays in sync with nightly ingestion pipelines. Branches provide an isolated sandbox for schema evolution, allowing engineers to test new transformations without risking production stability. Together, these tools create a repeatable ETL workflow—cloning source schemas, batching billing data, and scheduling refreshes—while keeping the operational footprint comparable to a single managed Postgres service.

As data volumes grow, Xata’s roadmap includes a seamless upgrade path to hybrid analytics. The pg_duckdb extension lets PostgreSQL offload heavy aggregations to DuckDB, preserving the familiar SQL interface while tapping a columnar engine optimized for analytical workloads. For organizations that eventually outgrow on‑prem PostgreSQL, exporting to lake formats such as Apache Iceberg or integrating with cloud‑native warehouses remains straightforward thanks to the standardized schema. By starting with a simple Postgres warehouse, companies gain immediate insight, maintain agility, and avoid premature optimization, positioning themselves to scale analytics infrastructure only when business demand justifies it.