Announcing Azure HorizonDB
Affan Dar, Vice President of Engineering, PostgreSQL at Microsoft Charles Feddersen, Partner Director of Program Management, PostgreSQL at Microsoft Today at Microsoft Ignite, we’re excited to unveil the preview of Azure HorizonDB, a fully managed Postgres-compatible database service designed to meet the needs of modern enterprise workloads. The cloud native architecture of Azure HorizonDB delivers highly scalable shared storage, elastic scale-out compute, and a tiered cache optimized for running cloud applications of any scale. Postgres is transforming industries worldwide and is emerging as the foundation of modern data solutions across all sectors at an unprecedented pace. For developers, it is the database of choice for building new applications with its rich set of extensions, open-source API, and expansive ecosystems of tools and libraries. At the same time, but at the opposite end of the workload spectrum, enterprises around the world are also increasingly turning to Postgres to modernize their existing applications. Azure HorizonDB is designed to support applications across the entire workload spectrum from the first line of code in a new app to the migration of large-scale, mission-critical solutions. Developers benefit from the robust Postgres ecosystem and seamless integration with Azure’s advanced AI capabilities, while enterprises can gain a secure, highly available, and performant cloud database to host their business applications. Whether you’re building from scratch or transforming legacy infrastructure, Azure HorizonDB empowers you to innovate and scale with confidence, today and into the future. Azure HorizonDB introduces new levels of performance and scalability to PostgreSQL. The scale-out compute architecture supports up to 3,072 vCores across primary and replica nodes, and the auto-scaling shared storage supports up to 128TB databases while providing sub-millisecond multi-zone commit latencies. This storage innovation enables Azure HorizonDB to deliver up to 3x more throughput when compared with open-source Postgres for transactional workloads. Azure HorizonDB is enterprise ready on day one. With native support for Entra ID, Private Endpoints, and data encryption, it provides compliance and security for sensitive data stored in the cloud. All data is replicated across availability zones by default and maintenance operations are transparent with near-zero downtime. Backups are fully automated, and integration with Azure Defender for Cloud provides additional protection for highly sensitive data. All up, Azure HorizonDB offers enterprise-grade security, compliance, and reliability, making it ready for business use today. Since the launch of ChatGPT, there has been an explosion of new AI apps being built, and Postgres has become the database of choice due in large part to its vector index support. Azure HorizonDB extends the AI capabilities of Postgres further with two key features. We are introducing advanced filtering capabilities to the DiskANN vector index which enable query predicate pushdowns directly into the vector similarity search. This provides significant performance and scalability improvements over pgvector HNSW while maintaining accuracy and is ideal for similarity search over transactional data in Postgres. The second feature is built-in AI model management that seamlessly integrates generative, embedding, and reranking models from Microsoft Foundry for developers to use in the database with zero configuration. In addition to enhanced vector indexing and simplified model management to build powerful new AI apps, we’re also pleased to announce the general availability of Microsoft’s PostgreSQL Extension for VS Code that provides the tooling for Postgres developers to maximize their productivity. Using this extension, GitHub Copilot is context aware of the Postgres database which means less prompting and higher quality answers, and in the Ignite release, we’ve added live monitoring with one-click GitHub Copilot debugging where Agent mode can launch directly from the performance monitoring dashboard to diagnose Postgres performance issues and guide users to a fix. Alpha Life Sciences are an existing Azure customers “I’m truly excited about how Azure HorizonDB empowers our AI development. Its seamless support for Vector DB, RAG, and Agentic AI allows us to build intelligent features directly on a reliable Postgres foundation. With Azure HorizonDB, I can focus on advancing AI capabilities instead of managing infrastructure complexities. It’s a smart, forward-looking solution that perfectly aligns with how we design and deliver AI-powered applications.” Pengcheng Xu, CTO Alpha Life Sciences For enterprises that are modernizing their applications to Postgres in the cloud, the security and availability of Azure HorizonDB make it an ideal platform. However, these migrations are often complex and time consuming for large legacy codebase conversions. To simplify this and reduce the risk, we’re pleased to announce the preview of GitHub Copilot powered Oracle migration built into the PostgreSQL Extension for VS Code. Built into VS Code, teams of engineers can work with GitHub Copilot to automate the end-to-end conversion of complex database code using rich code editing, version control, text authoring, and deployment in an integrated development environment. Azure HorizonDB is the next generation of fully managed, cloud native PostgreSQL database service. Built on the latest Azure infrastructure with state-of-the-art cloud architecture, Azure HorizonDB is ready to for the most demanding application workloads. In addition to our portfolio of managed Postgres services in Azure, Microsoft is deeply invested into the open source Postgres project and is one of the top corporate upstream contributors and sponsors for the PostgreSQL project, with 19 Postgres project contributors employed by Microsoft. As a hyperscale Postgres vendor, it’s critical to actively participate in the open-source project. It enables us to better support our customers down to the metal in Azure, and to contribute our learnings from running Postgres at scale back to the community. We’re committed to continuing our investment to push the Postgres project forward, and the team is already active in making contributions to Postgres 19 to be released in 2026. Ready to explore Azure HorizonDB? Azure HorizonDB is initially available in Central US, West US3, UK South and Australia East regions. Customers are invited to apply for early preview access to Azure HorizonDB and get hands-on experience with this new service. Participation is limited, apply now at aka.ms/PreviewHorizonDBPostgreSQL for the enterprise: scale, secure, simplify
This week at Microsoft Ignite, along with unveiling the new Azure HorizonDB cloud native database service, we’re announcing multiple improvements to our fully managed open-source Azure Database for PostgreSQL service, delivering significant advances in performance, analytics, security, and AI-assisted migration. Let’s walk through nine of the top Azure Database for PostgreSQL features and improvements we’re announcing at Microsoft Ignite 2025. Feature Highlights New Intel and AMD v6-series SKUs (Preview) Scale to multiple nodes with Elastic Clusters (GA) PostgreSQL 18 (GA) Realtime analytics with Fabric Mirroring (GA) Analytical queries inside PostgreSQL with the pg_duckdb extension (Preview) Adding Parquet to the azure_storage extension (GA) Meet compliance requirements with the credcheck, anon & ip4r extensions (GA) Integrated identity with Entra token-refresh libraries for Python AI-Assisted Oracle to PostgreSQL Migration Tool (Preview) Performance and scale New Intel and AMD v6 series SKUs (Preview) You can run your most demanding Postgres workloads on new Intel and AMD v6 General Purpose and Memory Optimized hardware SKUs, now availble in preview These SKUs deliver massive scale for high-performance OLTP, analytics and complex queries, with improved price performance and higher memory ceilings. AMD Confidential Compute v6 SKUs are also in Public Preview, enabling enhanced security for sensitive workloads while leveraging AMD’s advanced hardware capabilities. Here’s what you need to know: Processors: Powered by 5th Gen Intel® Xeon® processor (code-named Emerald Rapids) and AMD's fourth Generation EPYC™ 9004 processors Scale: VM size options scale up to 192 vCores and 1.8 TiB IO: Using the NVMe protocol for data disk access, IO is parallelized to the number of CPU cores and processed more efficiently, offering significant IO improvements Compute tier: Available in our General Purpose and Memory Optimized tiers. You can scale up to these new compute SKUs as needed with minimal downtime. Learn more: Here's a quick summary of the v6 SKUs we’re launching, with links to more information: Processor SKU Max vCores Max Mem Intel Ddsv6 192 768 GiB Edsv6 192 1.8 TiB AMD Dadsv6 96 384 GiB Eadsv6 96 672 GiB DCadsv6 96 386 GiB ECadsv6 96 672 GiB Scale to multiple nodes with Elastic clusters (GA) Elastic clusters are now generally available in Azure Database for PostgreSQL. Built on Citus open-source technology, elastic clusters bring the horizontal scaling of a distributed database to the enterprise features of Azure Database for PostgreSQL. Elastic clusters enable horizontal scaling of databases running across multiple server nodes in a “shared nothing” architecture. This is ideal for workloads with high-throughput and storage-intensive demands such as multi-tenant SaaS and IoT-based workloads. Elastic clusters come with all the enterprise-level capabilities that organizations rely upon in Azure Database for PostgreSQL, including high availability, read replicas, private networking, integrated security and connection pooling. Built-in sharding support at both row and schema level enables you to distribute your data across a cluster of compute resources and run queries in parallel, dramatically increasing throughput and capacity. Learn more: Elastic clusters in Azure Database for PostgreSQL PostgreSQL 18 (GA) When PostgreSQL 18 was released in September, we made a preview available on Azure on the same day. Now we’re announcing that PostgreSQL 18 is generally available on Azure Database for PostgreSQL, with full Major Version Upgrade (MVU) support, marking our fastest-ever turnaround from open-source release to managed service general availability. This release reinforces our commitment to delivering the latest PostgreSQL community innovations to Azure customers, so you can adopt the latest features, performance improvements, and security enhancements on a fully managed, production-ready platform without delay. Now you can: Deploy PostgreSQL 18 in all public Azure regions. Perform in-place major version upgrades to PG18 with no endpoint or connection string changes. Use Microsoft Entra ID authentication for secure, centralized identity management in all PG versions. Enable Query Store and Index Tuning for built-in performance insights and automated optimization. Leverage the 90+ Postgres extensions supported by Azure Database for PostgreSQL. PostgreSQL 18 also delivers major improvements under the hood, ranging from asynchronous I/O and enhanced vacuuming to improved indexing and partitioning, ensuring Azure continues to lead as the most performant, secure, and developer-friendly PostgreSQL managed service in the cloud. Learn more: PostgreSQL 18 open-source release announcement Supported versions of PostgreSQL in Azure Database for PostgreSQL Analytics Real-time analytics with Fabric Mirroring (GA) With Fabric mirroring in Azure Database for PostgreSQL, now generally available, you can run your Microsoft Fabric analytical workloads and capabilities on near-real-time replicated data, without impacting the performance of your production PostgreSQL databases, and at no extra cost. Mirroring in Fabric connects your operational and analytical platforms with continuous data replication from PostgreSQL to Fabric. Transactions are mirrored to Fabric in near real-time, enabling advanced analytics, machine learning, and reporting on live data sets without waiting for traditional batch ETL processes to complete. This approach eliminates the overhead of custom integrations or data pipelines. Production PostgreSQL servers can run mission-critical transactional workloads without being affected by surges in analytical queries and reporting. With our GA announcement Fabric mirroring is ready for production workloads, with secure networking (VNET integration and Private Endpoints supported), Entra ID authentication for centralized identity management, and support for high availability enabled servers, ensuring business continuity for mirroring sessions. Learn more: Mirroring Azure Database for PostgreSQL flexible server Adding Parquet support to the azure_storage extension (GA) In addition to mirroring data directly to Microsoft Fabric, there are many other scenarios that require moving operational data into data lakes for analytics or archival. The complexity of building and maintaining ETL pipelines can be expensive and time-consuming. Azure Database for PostgreSQL now natively supports Parquet via the azure_storage extension, enabling direct SQL-based read/write to Parquet files in Azure Storage. This makes it easy to import and export data in Postgres without external tools or scripts. Parquet is a popular columnar storage format often used in big data and analytics environments (like Spark and Azure Data Lake) because of its efficient compression and query performance for large datasets. Now you can use the azure_storage extension to can skip an entire step: just issue a SQL command to write to and query from a Parquet file in Azure Blob Storage. Learn more: Azure storage extension in Azure Database for PostgreSQL Analytical queries inside PostgreSQL with the pg_duckdb extension (Preview) DuckDB’s columnar engine excels at high performance scans, aggregations and joins over large tables, making it particularly well-suited for analytical queries. The pg_duckdb extension, now available in preview for Azure Database for PostgreSQL combines PostgreSQL’s transactional performance and reliability with DuckDB’s analytical speed for large datasets. Together pg_duckdb and PostgreSQL are an ideal combination for hybrid OLTP + OLAP environments where you need to run analytical queries directly in PostgreSQL without sacrificing performance., To see the pg_duckdb extension in action check out this demo video: https://aka.ms/pg_duckdb Learn more: pg_duckdb – PostgreSQL extension for DuckDB Security Meet compliance requirements with the credcheck, anon & ip4r extensions (GA) Operating in a regulated industry such as Finance, Healthcare and Government means negotiating compliance requirements like HIPAA and PCI-DSS, GDPR that include protection for personalized data and password complexity, expiration and reuse. This week the anon extension, previously in preview, is now generally available for Azure Database for PostgreSQL adding support for dynamic and static masking, anonymized exports, randomization and many other advanced masking techniques. We’ve also added GA support for the credcheck extension, which provides credential checks for usernames, and password complexity, including during user creation, password change and user renaming. This is particularly useful if your application is not using Entra ID and needs to rely on native PostgreSQL users and passwords. If you need to store and query IP ranges for scenarios like auditing, compliance, access control lists, intrusion detection and threat intelligence, another useful extension announced this week is the ip4r extension which provides a set of data types for IPv4 and IPv6 network addresses. Learn more: PostgreSQL Anonymizer credcheck – PostgreSQL username/password checks IP4R - IPv4/v6 and IPv4/v6 range index type for PostgreSQL The Azure team maintains an active pipeline of new PostgreSQL extensions to onboard and upgrade to Azure Database for PostgreSQL For example, another important extension upgraded this week is pg_squeeze which removes unused space from a table. The updated 1.9.1 version adds important stability improvements. Learn more: List of extensions and modules by name Integrated identity with Entra token-refresh libraries for Python In a modern cloud-connected enterprise, identity becomes the most important security perimeter. Azure Database for PostgreSQL is the only managed PostgreSQL service with full Entra integration, but coding applications to take care of Entra token refresh can be complex. This week we’re announcing a new Python library to simplify Entra token refresh. The library automatically refreshes authentication tokens before they expire, eliminating manual token handling and reducing connection failures. The new python_azure_pg_auth library provides seamless Azure Entra ID authentication and supports the latest psycopg and SQLAlchemy drivers with automatic token acquisition, validation, and refresh. Built-in connection pooling is available for both synchronous and asynchronous workloads. Designed for cross-platform use (Windows, Linux, macOS), the package features clean architecture and flexible installation options for different driver combinations. This is our first milestone in a roadmap to add token refresh for additional programming languages and frameworks. Learn more, with code samples to get started here: https://aka.ms/python-azure-pg-auth Migration AI-Assisted Oracle to PostgreSQL Migration Tool (Preview) Database migration is a challenging and time-consuming process, with multiple manual steps requiring schema and apps specific information. The growing popularity, maturity and low cost of PostgreSQL has led to a healthy demand for migration tooling to simplify these steps. The new AI-assisted Oracle Migration Tool preview announced this week greatly simplifies moving from Oracle databases to Azure Database for PostgreSQL. Available in the VS Code PostgreSQL extension the new migration tool combines GitHub Copilot, Azure OpenAI, and custom Language Model Tools to convert Oracle schema, database code and client applications into PostgreSQL-compatible formats. Unlike traditional migration tools that rely on static rules, Azure’s approach leverages Large Language Models (LLMs) and validates every change against a running Azure Database for PostgreSQL instance. This system not only translates syntax but also detects and fixes errors through iterative re-compilation, flagging any items that require human review. Application codebases like Spring Boot and other popular frameworks are refactored and converted. The system also understands context by querying the target Postgres instance for version and installed extensions. It can even invoke capabilities from other VS Code extensions to validate the converted code. The new AI-assisted workflow reduces risk, eliminates significant manual effort, and enables faster modernization while lowering costs. Learn more: https://aka.ms/pg-migration-tooling Be sure to follow the Microsoft Blog for PostgreSQL for regular updates from the Postgres on Azure team at Microsoft. We publish monthly recaps about new features in Azure Database for PostgreSQL, as well as an annual blog about what’s new in Postgres at Microsoft.Ignite 2025: Advancing Azure Database for MySQL with Powerful New Capabilities
At Ignite 2025, we’re introducing a wave of powerful new capabilities for Azure Database for MySQL, designed to help organizations modernize, scale, and innovate faster than ever before. From enhanced high availability and seamless serverless integrations to AI-powered insights and greater flexibility for developers, these advancements reflect our commitment to delivering a resilient, intelligent data platform. Join us as we unveil what’s next for MySQL on Azure - and discover how industry leaders are already building the future with confidence. Enhanced Failover Performance with Dedicated SLB for High-Availability Servers We’re excited to announce the General Availability of Dedicated Standard Load Balancer (SLB) for HA-enabled servers in Azure Database for MySQL. This enhancement introduces a dedicated SLB to High Availability configurations for servers created with public access or private link. By managing the MySQL data traffic path, SLB eliminates the need for DNS updates during failover, significantly reducing failover time. Previously, failover relied on DNS changes, which caused delays due to DNS TTL (30 seconds) and client-side DNS caching. What’s new with GA: The FQDN consistently resolves to the SLB IP address before and after failover. Load-balancing rules automatically route traffic to the active node. Removes DNS cache dependency, delivering faster failovers. Note: This feature is not supported for servers using private access with VNet integration. Learn more Build serverless, event-driven apps at scale – now GA with Trigger Bindings for Azure Functions We’re excited to announce the General Availability of Azure Database for MySQL Trigger bindings for Azure Functions, completing the full suite of Input, Output, and Trigger capabilities. This feature lets you build real-time, event-driven applications by automatically invoking Azure Functions when MySQL table rows are created, updated, or deleted - eliminating custom polling and boilerplate code. With native support across multiple languages, developers can now deliver responsive, serverless solutions that scale effortlessly and accelerate innovation. Learn more Enable AI agents to query Azure Database for MySQL using Azure MCP Server We’re excited to announce that Azure MCP Server now supports Azure Database for MySQL, enabling AI agents to query and manage MySQL data using natural language through the open Model Context Protocol (MCP). Instead of writing SQL, you can simply ask questions like “Show the number of new users signed up in the last week in appdb.users grouped by day.”, all secured with Microsoft Entra authentication for enterprise-grade security. This integration delivers a unified, secure interface for building intelligent, context-aware workflows across Azure services - accelerating insights and automation. Learn more Greater networking flexibility with Custom Port Support Custom port support for Azure Database for MySQL is now generally available, giving organizations the flexibility to configure a custom port (between 25001 and 26000) during new server creation. This enhancement streamlines integration with legacy applications, supports strict network security policies, and helps avoid port conflicts in complex environments. Supported across all network configurations - including public access, private access, and Private Link - custom port provisioning ensures every new MySQL server can be tailored to your needs. The managed experience remains seamless, with all administrative capabilities and integrations working as before. Learn more Streamline migrations and compatibility with Lower Case Table Names support Azure Database for MySQL now supports configuring lower_case_table_names server parameter during initial server creation for MySQL 8.0 and above, ensuring seamless alignment with your organization’s naming conventions. This setting is automatically inherited for restores and replicas, and cannot be modified. Key Benefits: Simplifies migrations by aligning naming conventions and reducing complexity. Enhances compatibility with legacy systems that depend on case-insensitive table names. Minimizes support dependency, enabling faster and smoother onboarding. Learn more Unlock New Capabilities with Private Preview Features at Ignite 2025 We’re excited to announce that you can now explore two powerful capabilities in early access - Reader Endpoint for seamless read scaling and Server Rename for greater flexibility in server management. Scale reads effortlessly with Reader Endpoint (Private Preview) We’re excited to announce that the Reader Endpoint feature for Azure Database for MySQL is now ready for private preview. Reader Endpoint provides a dedicated read-only endpoint for read replicas, enabling automatic connection-based load balancing of read-only traffic across multiple replicas. This simplifies application architecture by offering a single endpoint for read operations, improving scalability and fault tolerance. Azure Database for MySQL supports up to 10 read replicas per primary server. By routing read-only traffic through the reader endpoint, application teams can efficiently manage connections and optimize performance without handling individual replica endpoints. Reader endpoints continuously monitor the health of replicas and automatically exclude any replica that exceeds the configured replication lag threshold or becomes unavailable. To enroll in the preview, please submit your details using this form. Limitations During Private Preview: Only performance-based routing is supported in this preview. Certain settings such as routing method and the option to attach new replicas to the reader endpoint can only be configured at creation time. Only one reader endpoint can be created per replica group. Including the primary server as a fallback for read traffic when no replicas are available is not supported in this preview. Get flexibility in server management with Server Rename (Private Preview) We’re excited to announce the Private Preview of Server Rename for Azure Database for MySQL. This feature lets you update the name of an existing MySQL server without recreating it, migrating data, or disrupting applications - making it easier to adopt clear, consistent naming. It provides a near zero-downtime path to a new hostname of the server. To enroll in the preview, please submit your details using this form. Limitations During Private Preview: Primary server with read replicas: Renaming a primary server that has read replicas keeps replication healthy. However, the SHOW SLAVE STATUS output on the replicas will still display the old primary server's name. This is a display inconsistency only and does not affect replication. Renaming is currently unsupported for servers using Customer Managed Key (CMK) encryption or Microsoft Entra Authentication (Entra Id). Real-World Success: Azure Database for MySQL Powers Resilient Applications at Scale Factorial Factorial, a leading HR software provider, uses Azure Database for MySQL alongside Azure Kubernetes Service to deliver secure, scalable HR solutions for thousands of businesses worldwide. By leveraging Azure Database for MySQL’s reliability and seamless integration with cloud-native technologies, Factorial ensures high availability and rapid innovation for its customers. Learn more YES (Youth Employment Service) South Africa’s largest youth employment initiative, YES, operates at national scale by leveraging Azure Database for MySQL to deliver a resilient, centralized platform for real-time job matching, learning management, and career services - connecting thousands of young people and employers, and helping nearly 45 percent of participants secure permanent roles within six months. Learn more Nasdaq At Ignite 2025, Nasdaq will showcase how it uses Azure Database for MySQL - alongside Azure Database for PostgreSQL and other Azure products - to power a secure, resilient architecture that safeguards confidential data while unlocking new agentic AI capabilities. Learn more These examples demonstrate that Azure Database for MySQL is trusted by industry leaders to build resilient, scalable applications - empowering organizations to innovate and grow with confidence. We Value Your Feedback Azure Database for MySQL is built for scale, resilience, and performance - ready to support your most demanding workloads. With every update, we’re focused on simplifying development, migration, and management so you can build with confidence. Explore the latest features and enhancements to see how Azure Database for MySQL meets your data needs today and in the future. We welcome your feedback and invite you to share your experiences or suggestions at AskAzureDBforMySQL@service.microsoft.com Stay up to date by visiting What's new in Azure Database for MySQL, and follow us on YouTube | LinkedIn | X for ongoing updates. Thank you for choosing Azure Database for MySQL!Building brighter futures: How YES tackles youth unemployment with Azure Database for MySQL
YES leverages Azure Database for MySQL to power South Africa’s largest youth employment initiative, delivering scalable, reliable systems that connect thousands of young people to jobs and learning opportunities.Exciting things on the horizon for PostgreSQL fans @ Ignite 2025
If you’re passionate about PostgreSQL or just curious about what’s new, you’ll want to join us at Microsoft Ignite 2025. We have a packed lineup, including sessions exploring cutting-edge features and exclusive giveaways at the PostgreSQL on Azure booth. Haven’t registered yet? Now’s the time – sign up for Microsoft Ignite and start building your schedule. Below are the must-see PostgreSQL on Azure activities, with highlights of what you’ll learn at each. Add these to your agenda today. Sessions can fill up fast! Theater sessions: get a first look, fast I know from experience that attention spans can start to wane after hours-long keynotes, content-rich sessions, and conference socializing. Luckily, we have a couple of theater sessions that offer snackable but substantial information in less time than it will take to grab lunch. And they’re located conveniently on the main conference floor. PostgreSQL on Azure: Your launchpad for intelligent apps and agents (THR705) - See how we’re making PostgreSQL AI-aware for developers to drive app and agent innovation. Includes a demo of vector similarity search, semantic operators baked into Postgres, and more! Simplifying scale-out of PostgreSQL for performant multi-tenant apps (THR706) - Discover a smarter, simpler way to scale PostgreSQL using the new Elastic Clusters feature. If your app or service is growing fast (or you want it to!), add this breakout to learn how Azure makes it easier to scale Postgres and keep it reliable. These talks are a great way to sample what’s new and decide where to dive deeper. Plus, they’re fun and demo-heavy, and who doesn’t love a good demo? Breakout sessions: a deep dive into Postgres innovations Led by Azure product leaders and executives from organizations driving innovation backed by PostgreSQL, these breakout sessions will dive into the coolest new capabilities and real-world use cases. If you want rich, technical content and more live demos, these are for you. Build mission-critical apps that scale with PostgreSQL on Azure (BRK127) - Get a closer look at the next generation of PostgreSQL on Azure. Add this session, if you’re curious about how we’re taking Postgres to the next level to support your mission-critical AI workloads. Modern data, modern apps: Innovation with Microsoft Databases (BRK134) - Gain insider knowledge on the latest innovations across open-source, SQL, and NoSQL databases, and understand how Microsoft’s integrated database portfolio supports next-gen innovation. Nasdaq Boardvantage: AI-driven governance on PostgreSQL and AI Foundry (BRK137) - Discover how a Fortune 100 merges trust with cutting-edge AI leveraging Azure’s AI-enriched and enterprise-ready solutions, including Azure Database for PostgreSQL, Azure Database for MySQL, Azure AI Foundry, Azure Kubernetes Service (AKS), and API Management. AI-assisted migration: The path to powerful performance on PostgreSQL (BRK123) - A before and after migration journey from Oracle to Azure Database for PostgreSQL. See how the new AI-assisted migration experience delivers conversion in a few clicks and minimal downtime. The blueprint for intelligent AI agents backed by PostgreSQL (BRK130) - If you’re into AI development, this session will spark ideas on bridging the gap between raw data and AI reasoning. You’ll leave with practical tips to turbocharge your AI agents with PostgreSQL. Each breakout session is 45 minutes with live demos and Q&A, so you’ll get plenty of detail and interaction with Postgres experts. Hands-on lab: experience coding with Azure superpowers Do you learn best by doing? Then our guided workshop, Build advanced AI agents with PostgreSQL (Lab515), is for you. In each 75-minute session, you’ll get to create a fully functional AI-powered application backed by PostgreSQL on Azure with step-by-step guidance and expert insight on the latest innovations enabling intelligent app development. All the tools and instructions you’ll need are provided. Labs have limited capacity, so be sure to reserve your seat for any of the four labs in advance. This lab is a great way to understand how all the pieces come together on Azure. And you’ll gain practical skills you can apply to your own projects, whether it’s customer support bots, intelligent search in your app, or any scenario where PostgreSQL + AI collide. Expert meet-up booth: meet the team, grab some swag If you still want more Postgres (or a little Postgres souvenir), you can stop by the PostgreSQL on Azure Expert Meetup booth in the Ignite Hub. This will be our homebase on the show floor, where you can: Meet the team: I’ll be there in person, along with engineers, program managers, cloud solution architects, and advocates from our team. Whether you have a burning technical question, want to share feedback, or need guidance for your specific use case, come chat with us. Get a quick demo re-run: Sometimes a 5-minute demo is worth a thousand words, especially after you’ve sat through all those words already in a keynote. The booth will have a monitor and a live environment so we can walk you through select use cases if you have questions - no appointment needed. Swag and giveaways: Ah yes, the goodies! We know conference swag is part of the fun, so we’ve got some special PostgreSQL-themed giveaways at the booth. I won’t spoil all the surprises, but rumor has it there are some limited-edition items up for grabs. Network with peers: The expert meet-up area is also a magnet for PostgreSQL enthusiasts. You might bump into other attendees at the booth who are tackling similar projects or challenges. Ignite is about community as much as content, so come by and spark up a conversation. Meet you there? Ignite is our largest event of the year. We love sharing what we’ve been working on and, most of all, hearing from you, the community. So, on behalf of the Azure for PostgreSQL team, thank you for your interest and support. We can’t wait to show you what’s new and to help you continue to succeed with Postgres. See you in San Francisco!Azure PostgreSQL Lesson Learned #3: Fix FATAL: sorry, too many clients already
We encountered a support case involving Azure Database for PostgreSQL Flexible Server where the application started failing with connection errors. This blog explains the root cause, resolution steps, and best practices to prevent similar issues.Preventing and recovering from accidental deletion of an Azure Database for MySQL flexible server
Accidental deletion of critical Azure resources, such as Azure Database for MySQL flexible servers, can disrupt operations. To help avoid such accidental deletions, you can use a couple of options, including Azure Resource Locks and Azure Policy. This post explains how to implement these mechanisms, and how to revive a dropped MySQL flexible server by using the Azure CLI.Custom Port Support in Azure Database for MySQL – Flexible Server is Now Generally Available
We are excited to announce that custom port support for Azure Database for MySQL – Flexible Server is now generally available (GA). This long-requested feature gives you greater flexibility to align MySQL server deployments with your network and security requirements. By default, MySQL uses TCP port 3306; with this GA release, you can configure a custom port (between 25001 and 26000) when creating a new Azure Database for MySQL flexible server. This enables easier integration with legacy applications, helps comply with strict network security policies, and avoids port conflicts in complex environments. What’s new in GA (vs. Public Preview): In the Public Preview (July 2025), custom ports were only supported for VNet-injected (private access) servers, with no support for public access or Private Link connectivity. Now, with GA, you can create custom-port servers in any network configuration – including both publicly accessible servers and those using Private Link (private endpoint) connectivity. In short, all new MySQL flexible servers can be created with a custom port, whether they are configured for public network access or deployed into a private virtual network. Feature Highlights Custom Port Range: Choose a port between 25001 and 26000 during server provisioning. (Only one custom port is supported per server.) This is in addition to the default MySQL port 3306, which remains available for use if needed. Supported Scenarios: Custom ports are fully supported for new server creation, point-in-time restore (including cross-port restores), read replica setup, and High Availability (HA) deployments. You can perform a restore or set up a replica even if the source and target servers use different ports, and you can enable HA on a server configured with a non-default port. Networking Flexibility: Supported on both public access and private access configurations. You can create servers with a custom port in public access mode (accessible via the internet with firewall rules) or in private access mode (injected into a VNet). Azure Private Link is also supported – meaning you can connect via a private endpoint to a MySQL server running on a custom port. This enhancement broadens the feature’s applicability beyond the preview’s limited scope, allowing usage in all network scenarios. Managed Experience: The custom port feature is built into the managed service experience. Aside from specifying a different port number for client connections, there is no change in how you manage or operate the MySQL flexible server – all administrative capabilities and integrations (backup, monitoring, etc.) work as they do with the default port. Current Limitations Be aware of a couple of limitations at GA: Port Immutable After Creation: You cannot change the server’s port after the server is created. If you need to use a different port, you will have to create a new server with that port. As a workaround, you can use Point-in-Time Restore (PITR) to quickly clone your database into a new server with the desired port (since cross-port restores are supported), rather than performing a full manual migration. Geo-Replication/Geo-Restore: Cross-region operations like geo-restore and geo-replication are not yet supported for servers using a custom port. In other words, you cannot perform a geo-restore of a backup from a custom-port server, and you cannot create cross-region read replicas on custom port servers at this time. These capabilities are on the roadmap but remain unsupported in the current release. Why Custom Ports? Many enterprise developers and DBAs have asked for custom port support to accommodate specialized network scenarios. For example, some organizations enforce strict firewall rules or use non-standard ports for databases to meet internal security compliance requirements. Others may have legacy applications or multi-database setups that require MySQL to run on a port other than 3306 to avoid conflicts. The custom port feature addresses these needs by allowing you to select a non-default port during server creation, while Azure continues to handle all the usual PaaS management tasks. In short, you get the flexibility of a custom network configuration without losing the benefits of a fully managed database service. Getting Started Using a custom port is straightforward. At GA, the Azure portal’s create experience is the way to set a custom port (support in CLI/PowerShell/ARM will come later). In the portal, when you create a new Azure Database for MySQL – Flexible Server, you’ll find an option to specify the “Database port.” Provide any value between 25001 and 26000 as the port number for your server. Once the server is deployed, client applications should connect using the <servername>.mysql.database.azure.com hostname and the port you chose, instead of the default 3306. All other connection settings (such as SSL enforcement and credentials) remain the same. Make sure to configure network access rules to allow traffic on your chosen port. For public access servers, this means updating the firewall rules or network security groups to permit the custom port. For private access or Private Link setups, ensure that your networking (NSGs, on-premises firewall rules, etc.) permits traffic on the custom port to reach the database. Learn More Custom port support is now GA and ready for production use, so we encourage you to try it out if your environment can benefit from it. For more details on Azure Database for MySQL – Flexible Server connectivity and custom ports, refer to the official documentation: Networking Overview - Azure Database for MySQL | Microsoft Learn We look forward to seeing how you use this new capability to tailor your MySQL deployments. With custom port support now generally available, Azure Database for MySQL – Flexible Server offers even more flexibility to meet your organizational policies and integration needs, all while delivering a fully managed experience. Happy deploying!
