Introducing Lakehouse//RT: Real-Time Performance on a Unified Lakehouse

2026-06-16 · Source: Databricks · Field: Technology & Digital — Data Science & Analytics, Cloud Computing & IT Infrastructure, Artificial Intelligence & Machine Learning · Depth: Intermediate, long

Summary

Databricks has introduced Lakehouse//RT, a new real-time data warehouse designed for operational analytics, BI, and application serving workloads. Powered by the Reyden engine, Lakehouse//RT aims to unify real-time serving directly with the core lakehouse architecture, eliminating the need for separate serving layers that introduce data duplication, governance inconsistencies, and engineering overhead. The platform delivers millisecond performance at high concurrency, achieving sub-100ms latency at 12,000 queries per second on benchmarks. Preview participants reported up to 16x better performance compared to traditional real-time serving layers, with response times as low as 10ms. Lakehouse//RT maintains low latency under heavy load, at scale (up to a terabyte), and on complex queries, leveraging open data formats and centralizing governance via Unity Catalog. It is currently available in Beta for select read-only workloads, with an introductory offer of 30% off through January 2027.

Key takeaway

For AI Architects and Data Engineers evaluating real-time data architectures, Lakehouse//RT offers a compelling alternative to fragmented systems. You can achieve millisecond-latency performance for operational analytics and application serving directly on your lakehouse, eliminating the complexity, cost, and governance risks associated with maintaining separate serving layers. Consider migrating existing real-time workloads to Lakehouse//RT to streamline your data pipelines and ensure consistent governance across all data assets.

Key insights

Lakehouse//RT unifies real-time serving with the core lakehouse, eliminating separate layers for millisecond performance and simplified governance.

Principles

• Separate serving layers incur duplication, governance, and engineering costs.
• Real-time systems must perform under load, at scale, and on complex queries.
• Unified data platforms reduce complexity and operational overhead.

Method

Lakehouse//RT uses the Reyden engine with auto-sizing and incremental autoscaling to deliver real-time performance directly on open lakehouse data, without data movement.

In practice

• Consolidate operational analytics and BI onto a single platform.
• Leverage Unity Catalog for consistent governance across real-time data.
• Utilize auto-sizing and incremental autoscaling for cost-efficient compute.

Topics

• Lakehouse Architecture
• Real-time Data Warehousing
• Operational Analytics
• Databricks Lakehouse//RT
• Reyden Engine
• Unity Catalog
• Data Governance

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Data Engineer, AI Architect, MLOps Engineer

Related on AIssential

• How ERGO Hestia reduced time-to-market with Lakebase and Mosaic AI Model Serving — Consolidating data and model serving within a unified lakehouse platform accelerates innovation, simplifies operations, and enhances governance for real-time applications.
• Operational databases: How they work and when to use them — Modern data applications require unified operational and analytical capabilities beyond traditional OLTP systems.
• What is a Data Lakehouse? — A data lakehouse unifies data warehouse reliability and data lake scale on a single storage layer using open table formats.
• The Evolution of System Architecture Based on Need in Real-time Data Processes — System architecture must evolve from simple monoliths to complex hybrids to meet changing real-time data processing needs.
• Your Data, Your Lake: How Observe Uses Iceberg and Streaming ETL for Observability — Lakehouse architectures can provide scalable, cost-efficient observability by centralizing diverse telemetry data.
• "Skill issues'': data-centric optimization of lakehouse agents — A branching lakehouse enables data-centric optimization of coding agent skills by verifying induced state changes, improving accuracy by 31.9%.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Databricks.