• Cloud Database Insider
  • Posts
  • Google Launches Data Cloud for Agentic AI☁️|Mercedes-Benz pioneers cross-cloud data mesh 🚗🌐|Databricks Lakeflow vs Airflow⚙️

Google Launches Data Cloud for Agentic AI☁️|Mercedes-Benz pioneers cross-cloud data mesh 🚗🌐|Databricks Lakeflow vs Airflow⚙️

Deep Dive: My time at the Fabric SQL Future Ready Roadshow

Author

Gladstone Benjamin
April 27, 2026

In partnership with

Databricks Lakeflow vs Airflow: Workflow orchestration showdown ⚙️

Also, check out the weekly Deep Dive - Fabric SQL Future Ready Roadshow

Analytics on Live Data Without Leaving Postgres

When analytics on Postgres slows down, most teams add a second database. TimescaleDB by Tiger Data takes a different approach: extend Postgres with columnar storage and time-series primitives to run analytics on live data, no split architecture, no pipeline lag, no new query language to learn. Start building for free. No credit card required.

GCP

TL;DR: Google launched a data cloud tailored for autonomous agentic AI, integrating AI into cloud infrastructure to enable scalable, secure real-time analytics and boost AI-driven automation across industries.

  • Google launched a data cloud platform specifically designed to support autonomous, agentic AI systems efficiently.

  • The platform integrates AI capabilities with cloud infrastructure, enhancing data accessibility, governance, and security.

  • It supports scalable, flexible AI model deployment, real-time analytics, and improved decision-making across workloads.

  • This innovation accelerates AI adoption in industries and highlights competitive trends among cloud service providers.

Why this matters: Google's data cloud tailored for agentic AI empowers businesses to deploy autonomous AI with greater scalability, security, and real-time insight. This innovation accelerates AI integration across industries, setting new standards for cloud services and driving competitive advances in intelligent automation and ethical data governance.

DATA ARCHITECTURE

TL;DR: Mercedes-Benz built a cross-cloud data mesh using Delta Sharing and intelligent replication for real-time, cost-effective data sharing, boosting agility, collaboration, and digital transformation across business units and partners.

  • Mercedes-Benz created a cross-cloud data mesh using Databricks' Delta Sharing and intelligent replication technologies.

  • Delta Sharing enables real-time, open data sharing across clouds without copying large data sets, ensuring consistent freshness.

  • Intelligent replication selectively distributes critical data to reduce latency and cloud costs, optimizing data accessibility.

  • This strategy enhances agility, collaboration, and digital transformation while setting new industry standards for multi-cloud data use.

Why this matters: Mercedes-Benz's innovative cross-cloud data mesh using Delta Sharing and intelligent replication drives efficient, secure, real-time data collaboration across clouds. This boosts agility, cuts costs, and accelerates digital transformation, setting a benchmark for scalable, decentralized data architectures in the automotive industry and beyond.

DATA ENGINEERING

TL;DR: Databricks Lakeflow Jobs offer seamless, scalable automation tightly integrated with Delta Lake, while Airflow provides flexible, multi-tool orchestration but demands more manual setup and management. Choose based on infrastructure and workflow needs.

  • Databricks Lakeflow Jobs integrate deeply with Lakehouse and Delta Lake, offering scalable, cloud-native workflow automation.

  • Airflow provides flexible, complex DAG orchestration with broad community support but requires more manual setup.

  • Lakeflow Jobs suit teams heavily using Databricks, reducing management overhead with built-in monitoring and lineage features.

  • Airflow benefits organizations needing multi-tool orchestration and customizable workflows across diverse data toolchains.

Why this matters: Choosing between Databricks Lakeflow Jobs and Airflow can dramatically affect data pipeline efficiency, scalability, and maintenance. Lakeflow Jobs offer streamlined integration for Databricks-centric teams, while Airflow’s flexibility suits complex, multi-tool environments, highlighting the need to align orchestration tools with organizational infrastructure and expertise.

EVERYTHING ELSE IN CLOUD DATABASES

DEEP DIVE

Fabric SQL Future Ready Roadshow

It’s always a great experience to visit Microsoft Canada HQ and attend a periodic, one day conference. In this case it was the Fabric SQL Future Ready Roadshow. There are a some pretty nice views from there too:

I also met a couple of former colleagues there and networked as well. A good by-product of getting out of the office.

I learned a few new things, in particular, the expanded feature set of SQL Server 2025.

Anyhow, it was an excellent day of learning.

The following are the key points from the Fabric SQL Future Ready Roadshow with some images from the slides of the presentations:

OneLake & Data Unification OneLake is Microsoft's unified data lake within Fabric. It supports three ways to bring data in: Move (batch copy from anything), Shortcut (virtualize connections to file storage like Databricks, ADLS, SharePoint, AWS S3, GCS, Snowflake), and Mirror (real-time replication from databases like MySQL, Google BigQuery, Snowflake, Oracle, SAP, SQL Server, Cosmos DB). The goal is zero-ETL data unification — shortcuts virtualize to data storage while mirroring replicates from databases, and mirroring is free. New shortcut/mirroring sources in public preview include SharePoint+OneDrive, Azure Database for MySQL, and Google BigQuery, with Dremio and Azure Monitor coming soon. Snowflake and Databricks Catalog are getting native integration.

Mirroring Deep Dive Mirroring lets you replicate data from sources across Azure, on-prem, and multi-cloud into Fabric's compute engines (Spark, T-SQL, KQL, Analysis Services, Power BI) — all stored in OneLake in Delta format. The Databricks Catalog mirror is notable: only the catalog metadata is mirrored, not the data itself. Shortcuts are created per table, data stays in sync with no staleness, and Fabric/OneLake reads directly from ADLS with delegated identity tokens.

Azure SQL & SQL Server Capabilities Azure SQL is positioned as "the developer's database" with features spanning JSON and vector data types, regex/fuzzy matching, multi-model support (graph, spatial, columnar, document, key-value, hierarchical, in-memory, ledger), change event streaming/CDC/CT, embeddings generation, a Data API builder (DAB) for REST & GraphQL, SQL MCP Server, full-text/BM25/DiskANN search, and native support for Aspire, EF Core, Semantic Kernel, and Agent Framework.

Mission-Critical Scale Azure SQL supports up to 128 TB databases (auto-scaling from 10 GB), 192 vCores, 30+ named read replicas, 99.995% availability SLA with zone redundancy, and 150 MiB/s transaction log throughput independent of compute. The cloud architecture separates compute and storage, uses paired page servers with RBPEX SSD cache, an externalized log service, and fast backup/restore via storage snapshots.

AI & Agents Data Agents use an NL2SQL flow — a user asks a natural-language question against a sales database, the agent invokes an NL2SQL tool, determines available context and which tools to use, then creates a message with the answer. Database Agents are a new capability for monitoring and tuning entire database fleets, built around three pillars: guided actions (signals to decisions), built-in governance (enterprise-grade trust), and native tools (always-on assistant).

AI Apps Announcements All in public preview: SQL MCP Server via Data API Builder (DAB) 2.0, DML on vector-indexed tables, iterative filtering for vector indexes with TOP APPROXIMATE, and improved CREATE VECTOR INDEX performance.

Database Hub in Fabric A new unified management plane announced for Fabric that brings together Azure Database for MySQL, PostgreSQL, HorizonDB, Document DB, Azure SQL, SQL Server 2025, Cosmos DB, and Fabric Databases — with fleet management, observability, and database agents.

What's New Recap (SQLCon 2026) A dense summary slide organized into three tracks: Migrate & Modernize (Arc-based migration, SSMA Sybase-to-SQL with Copilot, Fabric SQL Database migration, managed instance link improvements, Linux support including Ubuntu 24.04/RHEL 10), Cloud-Native (ADX monitoring, AI features, security with TDE versionless keys and Entra principals, performance with 160/192 vCore support and automatic index compaction), and Unify Your Data Platform (SQL database in Fabric enterprise readiness, all collations, full-text search, dynamic data masking, mirroring of 500+ tables). Tools & SDKs updates include VS Code MSSQL extension with Copilot, SSMS 22 with Copilot, and updated drivers for Python, PHP 8.4/8.5, SqlClient 7.0, JDBC 13.4, and Django 6.0.

Azure Support Programs Microsoft highlighted a six-step support journey: AI Readiness Assessment, Envisioning Workshops, Solution Accelerators, Azure Innovate Program, FastTrack Support, and AI Landing Zone. The Azure Accelerate program offers access to partner experts, Microsoft investment funding/credits, free training, and AI-enhanced assessment tools across all industries and project stages.

Gladstone Benjamin

🚀 Work With Cloud Database Insider

Looking to reach enterprise data engineers and architects?

Limited sponsorship slots available each month.