BlueOnyx
DataDatabasesAnalyticsArchitectureOpen SourceAcquisition

The Dual-Stack Data Architecture Is Starting to Show Its Limits

Théodore BaillyPublished on 8 novembre 20265 min read
Présentation de données sur ordinateur portable à des collègues

Introduction

For years, data teams have operated two parallel worlds: transactional databases to run applications, and analytics engines to query historical records. EDB, the leading commercial provider of Postgres, announced on November 5, 2026 the acquisition of Altinity, the reference player for ClickHouse enterprise support and managed services. The stated goal: unify both worlds on a single platform without compromising the performance of either.

Two Complementary Engines, One Persistent Problem

ClickHouse is an open-source columnar analytics engine that has become a benchmark for real-time analysis at massive scale. Altinity, founded in 2017, built its reputation on enterprise-grade production deployments, producing tools that have since become community standards: a Kubernetes operator with over 108 million downloads on Docker Hub, stable builds, the Altinity.Cloud managed service, and the Antalya project for real-time data lakes. Organizations such as The Trade Desk, Tubi, and AppsFlyer are among its customer base.

Postgres, for its part, remains the default transactional database across a large share of the enterprise application stack. EDB has industrialized it through its EDB Postgres AI offering, covering on-premises, cloud, and far-edge deployments.

Bringing the two together addresses a classic architectural pain point: Postgres excels at OLTP workloads, frequent writes, transactional consistency, while ClickHouse handles massive analytical aggregations (OLAP). Running them separately means synchronization pipelines to maintain, duplicated data, and governance split across two independent systems.

What the Integration Will Actually Deliver

EDB is targeting delivery of integrated capabilities in Q4 2026. The technical centerpiece is a shared Apache Iceberg catalog, allowing both engines to read and write to the same open table format without physically duplicating data. Unified access controls will complement the integration, consolidating governance that currently sits fragmented across two independent systems.

Altinity, rebranded as the "EDB Real-Time Analytics" group, retains its leadership team and full product portfolio. Existing contracts, SLAs, and services for the combined base of approximately 1,860 enterprise customers remain unchanged in the near term. EDB also commits to keeping ClickHouse under the Apache 2.0 license, preserving deployment freedom for organizations that have built their stack on open source.

An Architectural Pressure That Keeps Building

This acquisition reflects a broader industry shift. By 2026, several major analytics vendors were already offering native Postgres services or zero-ETL integrations, a clear sign that the boundary between OLTP and OLAP has become a strategic friction point. The rise of AI agents is sharpening that pressure: an agent reasoning continuously over live data has little tolerance for pipelines with several minutes of latency between the transactional source and the analytics warehouse.

What This Means for Data Teams

For data engineers, the promise is a shared catalog that cuts redundant ETL pipelines and simplifies data lineage. For CIOs, the consolidation streamlines the vendor portfolio: one commercial relationship, a unified support plan, centralized access governance.

The integration still needs to prove itself in real-world conditions, the deliverables slated for late 2026 will be the true test of how cohesively the two engines actually work together. But the signal is unambiguous: the dual-stack data architecture, long accepted as an unavoidable cost of running serious analytics, is now being openly challenged by the vendors themselves.

Share

The Dual-Stack Data Architecture Is Starting to Show Its Limits