Lakebase

22 posts · Apr 2024 – Oct 2026

What this trail covers

Lakebase Postgres offers branch-based restores, enabling near-instant recovery of large datasets (e.g., 100 TB in seconds) by leveraging its architecture to replace traditional slow OLTP restore paths. This significantly improves disaster recovery and testing scenarios. This post introduces the LTAP (Lake Transactional/Analytical Processing) architecture, which enables loading terabytes of data into Lakebase Postgres in minutes. It details the safe and parallel construction of valid Postgres pages, and now integrates with Replit for enterprise application development, enabling governed development workflows.

All Databricks trails

Oct 2026Load terabytes of data in minutes into Lakebase PostgresThis post introduces the LTAP (Lake Transactional/Analytical Processing) architecture for Lakebase Postgres. It details the safe and parallel construction of valid Postgres pages, enabling the loading of terabytes of data in minutes. This architecture improves bulk load performance and enhances the safety of OLTP workloads.
Oct 2026How to build governed enterprise apps on Databricks with Replit and LakebaseThis post details the integration of Replit, a cloud-based IDE, with Databricks Lakebase. It provides a step-by-step guide for setting up this integration, enabling developers to build and deploy governed enterprise applications. The technical contribution lies in demonstrating how Replit's development environment can be connected to Lakebase for application development, leveraging Lakebase's capabilities for data persistence and management within a governed framework. This facilitates a more streamlined development-to-deployment pipeline for enterprise applications built on Databricks.
Oct 2026Lakebase Postgres branch-based restores for fast recovery at scaleThis post details how Lakebase Postgres implements branch-based restores, a novel approach to database recovery that replaces traditional slow OLTP restore paths. It explains how the architecture enables near-instantaneous recovery of large datasets (100 TB in seconds) by creating and managing branches, contrasting this with the limitations of traditional methods.
Sep 2026A practical guide to cost optimization with Lakebase PostgresThis post provides practical guidance on cost optimization for Lakebase Postgres. It elaborates on how Lakebase's separated storage and compute architecture enables cost-efficient use of branching, read replicas, and high availability. It also details methods for identifying and analyzing Lakebase costs through system tables, offering actionable tips for users to manage their spending.
Sep 2026Lakebase Search: State-of-the-art full text and vector search for PostgresThis post introduces Lakebase Search, which extends the Lakebase capability by integrating full-text and vector search directly into Postgres. This involves developing new indexing strategies and query processing mechanisms within Postgres to support both keyword-based and semantic search. The technical contribution lies in the implementation of these search functionalities as a native extension to Postgres, managed and orchestrated through the Lakebase framework, enabling advanced data retrieval for AI agents and other applications.
Advertisement
Sep 2026How I built agent-based security reviews on DatabricksThis post details the creation of an agent-based security review system on Databricks. The author describes building custom agents that leverage Databricks' capabilities, including Lakebase Postgres, to perform automated security reviews. The system involves defining security policies, orchestrating agents to execute these policies against code repositories, and analyzing the results to identify potential vulnerabilities. The technical depth includes discussions on agent design, policy definition, and the integration with Databricks services for execution and reporting.
Sep 2026Database for AI Agents: 5 Evaluation CriteriaThis post evaluates five criteria for selecting a database for AI agents: cost, latency, data freshness, scalability, and ease of integration. It discusses how Databricks Lakehouse, particularly when integrated with Postgres, can meet these criteria by providing a unified platform for data management and AI agent orchestration. The post emphasizes the importance of a robust data foundation for efficient and effective AI agent deployment.
Sep 2026Managed Postgres: What Lakebase Actually Takes Off Your PlateThis post details how Databricks' Managed Postgres offering, specifically the Lakebase capability, offloads operational burdens from users. It highlights the technical aspects of managing Postgres instances, including patching, upgrades, backups, and high availability, emphasizing the reduction in operational overhead for users. The post contrasts this with self-managed solutions, underscoring the engineering effort involved in providing a robust and reliable managed service.
Sep 2026Improving Lakebase Postgres compute cacheThis post details improvements to the compute cache for Lakebase Postgres, focusing on enhancing performance and efficiency. It likely involves architectural adjustments, caching strategies, or optimizations within the Lakehouse to accelerate data processing and retrieval for AI agents and other workloads leveraging Lakebase Postgres.
Sep 2026Build durable agents with Temporal and LakebaseThis post details the integration of Temporal for durable agent execution with Lakebase Postgres, enhancing the reliability and robustness of AI agents. It likely involves architectural patterns for state management, error handling, and recovery mechanisms when running agents on the Databricks platform, specifically leveraging Temporal's workflow orchestration capabilities alongside Lakebase's data storage and compute.
Sep 2026The 40-year-old database rule agents just broke: How LTAP unifies OLTP and OLAP workloadsIntroduces LTAP (Long-Term Archival Persistence) which unifies OLTP and OLAP workloads by enabling OLTP databases to query OLAP data directly. This breaks the traditional separation of these workloads by providing a unified data plane for low-latency OLTP access to OLAP stores. The system allows for direct querying of Databricks Lakehouse data from OLTP systems, enhancing performance and efficiency for AI agents and other workloads.
Aug 2026Vertical Advantage: Transforming Industries with Lakebase and Agentic AIThis post details how Databricks' Lakebase and Agentic AI capabilities are being used to transform industries. It highlights the synergy between unified data access (Lakebase) and intelligent automation (Agentic AI) to enable new applications and workflows. The post likely discusses how Lakebase's ability to unify OLTP and OLAP workloads, as described in previous posts, provides the foundational data layer for sophisticated agentic AI applications. It may also touch upon specific use cases or architectural patterns that demonstrate this integration, showcasing how agents can leverage the unified data plane for complex decision-making and task execution across various industries.
Aug 2026Object Storage + WAL: Lakebase Postgres for the agentic eraThis post details the integration of Postgres Write-Ahead Logging (WAL) with object storage to enable a unified OLTP and OLAP data plane, specifically for the 'agentic era'. It describes how WAL files are streamed to object storage, allowing for near real-time querying of transactional data alongside analytical data in the Lakehouse. This addresses the challenge of providing low-latency access to transactional data for AI agents and other applications that require up-to-date information without the overhead of traditional replication or ETL.
Aug 2026Ingest semi-structured data faster and more efficiently with Variant - Now Generally AvailableIntroduces the 'Variant' data type for semi-structured data ingestion, improving performance and efficiency within the Lakebase system.
Jul 2026Backstage with Lakebase, part 3This post, part 3 of the 'Backstage with Lakebase' series, likely delves into the internal architecture, operational challenges, or specific implementation details of the Lakebase system. It builds upon the foundational concept of unifying OLTP and OLAP workloads by enabling direct querying of OLAP data from OLTP databases. The 'Backstage' aspect suggests a focus on the engineering efforts behind Lakebase, potentially covering aspects like system design, scaling, migration, debugging, or operational management of the unified data plane.
Jul 2026Convert proprietary code to open ANSI SQL with Genie CodeIntroduces Genie Code, a tool that converts proprietary code to open ANSI SQL, enhancing interoperability and data accessibility within Databricks Lakebase.
Jul 2026NBCUniversal’s Seamless Migration: Unlocking Scalable Analytics with DatabricksThis post describes NBCUniversal's migration of their analytics platform from an on-premises Hadoop cluster to Databricks Lakebase. It highlights the technical challenges and solutions involved in scaling their analytics capabilities, including the adoption of Databricks SQL for improved performance and cost-effectiveness. The migration process involved re-architecting data pipelines and leveraging Delta Lake for data reliability.
Jul 2026Simplify AI agent orchestration with Lakebase PostgresThis post introduces the simplification of AI agent orchestration by leveraging Lakebase Postgres. It details how the combination of Databricks Lakehouse and Postgres can be used to manage and orchestrate AI agents, likely by providing a unified data plane and transactional capabilities for agent state, configuration, or execution. This contributes to the Lakebase capability by showcasing its application in a new domain (AI agent orchestration) and highlighting the technical benefits of using Postgres within the Lakehouse for this purpose.
Jul 2026Why R&D Data Belongs in the Lakehouse - and Why Agents Need It ThereThis post argues for the necessity of R&D data residing in the Lakehouse to effectively power AI agents. It details how agents require access to this data for tasks like knowledge retrieval, reasoning, and decision-making, and how the Lakehouse provides the necessary infrastructure for managing and serving this data at scale. The post emphasizes the technical benefits of having R&D data in the Lakehouse for agent development and deployment, including improved data quality, accessibility, and integration capabilities.
Apr 2024Granular Usage Attribution for dbt Pipelines with Query Tags - ClonedThis post introduces the use of query tags in Databricks to achieve granular usage attribution for dbt pipelines. It explains how these tags are applied to queries executed by dbt, allowing for the association of compute and storage costs with specific dbt models or jobs. The technical details likely involve how Databricks' query engine processes and reports on these tags, and how this data is integrated into the Lakebase system for billing and analysis.