Azure Weekly
Issue 574
9th August 2026
Highlights this week include:
- Secure by default: Trusted Launch as Default is now Generally Available by ehassan - Trusted Launch now auto-secures new Azure Gen 2 VMs with Secure Boot and virtual TPM after single subscription registration for various deployment methods.
- Marine Protected Areas in Microsoft Fabric Maps by Sander van de Velde - Fabric Maps leverages real-time IoT data to overlay marine protected areas in the Mediterranean Sea, integrating predictive insights on sea mammal protection alongside remote monitoring capabilities.
- ChainDrop supply chain compromise: Anatomy of a self-propagating worm by Microsoft Security Research, Ravikant Tiwari, Sagar Patil, Suriyaraj Natarajan - ChainDrop exposes a sophisticated self-propagating worm using obfuscated Bun JavaScript payloads to steal credentials from developer workstations and CI/CD pipelines for authentication across npm, GitHub, AWS, Kubernetes, and HashiCorp Vault, then auto-republishs compromised packages.
- Azure Key Vault Premium and AES-KW Post Quantum Resistant Key Wrapping in Public Preview by Michael Howard - Azure Key Vault introduces AES-KW for quantum-resistant key wrapping in public preview, allowing services to upgrade from RSA-based management.
- Trying Out WSL Containers: Microsoft's Built-In Alternative to Docker Desktop by Mike Evans-Larah - WSL Containers provides a built-in method to run Linux containers on Windows with Docker-CLI compatibility, though its current limitations-slow performance for Windows file system repos and lack of support for Docker Compose and .NET Aspire-mean it is promising but not yet ready to fully replace Docker Desktop.
- Turn one giant AI-generated pull request to a reviewable stack by Julia Muiruri - GitHub's new stacked pull requests break down large AI-generated code into smaller reviewable layers, enhancing collaboration and reducing merge conflicts for complex features.
- Your Cloud AI SLA Tells You Less Than You Think by James Dawson - Cloud AI service SLAs often overlook how frequent, unannounced model updates can silently degrade performance without triggering downtime alerts.
🤖 AI
- Learn how to use the four IQs: Web IQ, Work IQ, Fabric IQ, Foundry IQ The series shows how to use Work IQ in Python to retrieve user-specific M365 data like Teams chats, emails, and calendar events for AI application development.
- More powerful agents and workflows for autonomous business processes: Introducing a new harness for Copilot Studio The GitHub Copilot harness in Copilot Studio uses advanced models like Opus 5 and GPT-5.6 Sol to enhance complex business process agents, improving multi-tool use, file analysis, code analysis, and knowledge quality for enterprise workflows.
- Your Cloud AI SLA Tells You Less Than You Think Cloud AI service SLAs often overlook how frequent, unannounced model updates can silently degrade performance without triggering downtime alerts.
- From Build to Run to Distribute: Autonomous Agents with Microsoft Foundry Agent Service Part 1/5 The new GA release integrates the Foundry Agent Service with GitHub Copilot and the Copilot SDK for building running and distributing autonomous agents across Teams and 365 Copilot, referencing the open-source FibreOps outage response system.
- From SharePoint to Vector DB: How Enterprise RAG Ingestion Actually Works The article details the four-stage ingestion pipeline-Extract, Chunk, Embed, Store-that supports effective enterprise RAG systems, highlighting that issues with chunking and ingestion frequently lead to production failures more than model limitations.
- When Does RL Actually Help Fine-Tuning? A Difficulty-Controlled Study on Structured Generation Reinforcement learning provides only a small absolute improvement (+0.0111) for straightforward tasks but can boost performance by about 7.4% relative (+0.0515) on more difficult formulations where attributes must be inferred from prose, especially in Semantic-IoU-scored fields.
- Introducing Real-Time Multichannel Speech-to-Text Real-Time Multichannel Speech-to-Text in Azure Speech preserves stereo audio channel separation throughout transcription, simplifying contact center intelligence and agent-assist solution development by maintaining speaker attribution and conversational structure without extra infrastructure.
- Long-term memory in Microsoft Agent Framework The Microsoft Agent Framework adds long-term memory for AI agents to remember user interactions and preferences across sessions, using tools like ContextProviders and databases such as Neo4j or Azure Cosmos DB.
- Exploring Multi-Agent Workflows with Microsoft Agent Framework The Microsoft Agent Framework introduces five innovative multi-agent workflow patterns enabling organizations to design complex collaborative AI systems for tasks like customer support and content generation with improved accuracy and scalability.
- Beyond Model Evaluation: Choosing Between Microsoft Foundry and PyRIT for AI Red Teaming Choose Microsoft Foundry for model quality and safety evaluation, and use PyRIT for detailed red teaming on prompt injection, data leakage, and tool misuse because of their complementary strengths.
- Building a Conversational Alexa Skill with Microsoft Foundry Agents Building a Conversational Alexa Skill with Microsoft Foundry Agents shows how to integrate Alexa's voice interface with Foundry's reasoning and context features, addressing latency issues via Azure Functions for auth and protocol changes.
- From Distillation to Reinforcement Fine-Tuning: Hill-Climbing in Microsoft Foundry In Microsoft Foundry, reinforcement fine-tuning with a deterministic Python grader improved planning quality and reduced token cost per scenario while achieving 100% feasibility in 150 supply-chain tests.
- Microsoft Foundry's New Model Wave Microsoft Foundry now offers a diverse model portfolio including GPT-5.6 for advanced reasoning and Kimi-K2.7-Code for coding allowing enterprises to route workloads based on cost latency context size and modality requirements.
- Microsoft Foundry vs Claude: when should a business actually make the move? Use Claude for personal tasks and switch to Microsoft Foundry for business processes needing data access, accountability, and governance compliance.
- From Good to Great: We Put Agent Optimizer to the Test in Microsoft Foundry Agent Optimizer in Microsoft Foundry automates hand-tuned improvements by analyzing natural-language traces of agent failures, proposing targeted configuration changes to boost performance without full retraining costs or risks.
- The Hidden Reason Your Foundry Agent Can't Reach Any of Your Private (Bring-Your-Own) Resources The failure often stems from misconfigured capability hosts at both the Foundry account and project levels.
- Microsoft Agent Framework: The Complete Series The Microsoft Agent Framework series guides developers from basics to advanced topics including function tools, memory integration, human-in-the-loop approval, MCP support, and RAG, ending with a free video course and complete source code.
- Build a Local RAG Application in C# with Ollama and Qdrant – Part 1 This tutorial shows how to build a local RAG application in C# using Ollama for AI inference and Qdrant as the vector database, emphasizing privacy and efficiency by handling data and processing locally.
🔎 Analytics
- Generally Available: Unity AI Gateway on Azure Databricks Unity AI Gateway's GA on Azure Databricks provides centralized governance for AI workloads, enhancing model security, cost control, and compliance via unified monitoring and access management.
- Generally Available: SharePoint Connector for Azure Databricks The new GA SharePoint connector for Azure Databricks allows easy ingestion of SharePoint files into Azure Databricks through Lakeflow Connect, simplifying enterprise content integration with data and AI workflows.
- Announcing: Azure Databricks Genie One and Genie Agents Free Usage Extended Through January 31, 2027 Azure Databricks has extended free usage for Genie One and Genie Agents through January 31, 2027, removing budget restrictions during this promotional period.
- What is Fabric Apps? Fabric Apps is a platform that enables developers to build and deploy multi-cloud applications with unified management, streamlining operations across diverse cloud environments.
- Optimising DAX: The Formula Engine and Storage Engine Understanding how Power BI's Formula Engine handles complex calculations while the Storage Engine efficiently processes data is key to optimizing DAX query performance.
- Microsoft Fabric OneLake Shortcuts: When to Use Them and When Not To Fabric OneLake shortcuts enable Executive Analytics to integrate remote tables from various workspaces into a lakehouse without duplicating data, using hidden path references for seamless querying.
🖥️ Compute
- Generally Available: Trusted Launch as Default Azure’s Trusted Launch as Default now automatically enforces Secure Boot and vTPM on all new Gen2 VMs and scale sets, simplifying compliance with industry security standards.
- Retirement: Nested confidential (cc_v5) VMs will be retired on September 1, 2026 The cc_v5 confidential VM series, including the DCas_c size, will be discontinued on September 1, 2026, requiring users to migrate or resize their instances beforehand.
- Secure by default: Trusted Launch as Default is now Generally Available Trusted Launch now auto-secures new Azure Gen 2 VMs with Secure Boot and virtual TPM after single subscription registration for various deployment methods.
🚢 Containers
- Announcing Azure Linux on WSL - Beta Introducing Azure Linux on WSL enables developers to use the same cloud-optimized distribution across their workstation, CI pipelines, and Azure services, reducing environment drift and streamlining the transition from code to production.
- Cloud-Native Multi-Agent: Running Agents Safely on Kubernetes with Kars Kars secures enterprises by running multi-agent services as microservices on Kubernetes with strict network policies and audit trails.
- Building Execution Ledger - Durable Workflow Orchestration in Rust on Azure Container Apps Execution Ledger is an open-source Rust tool on Azure Container Apps that provides detailed step-level replay and compensation for robust long-running workflows, overcoming issues found in platforms such as Azure Durable Functions and Temporal by prioritizing recovery as a core feature.
- AKS Automatic Pod Disruption Budget Management Is Now in Preview Azure Kubernetes Service now offers preview support for Automatic Pod Disruption Budget management, which helps prevent AKS upgrades from stalling due to pod eviction restrictions.
- K8s DevOps AI Agent: AI-Powered KUBERNETES Operations The K8s DevOps AI Agent automates Kubernetes troubleshooting by integrating Azure Kubernetes Service, Azure OpenAI, and a React-based SRE dashboard to provide natural language incident analysis, root cause reports, and automated remediation recommendations in seconds.
🗄️ Databases
- Generally Available: Immutability to the most recent seven days of backups on Azure SQL Database and Azure SQL Managed Instance Azure SQL Database and Azure SQL Managed Instance now automatically enforce immutability on the latest seven days of backups, ensuring data protection without additional configuration.
- How to build an Oracle AWR-equivalent diagnostics stack in PostgreSQL (step-by-step guide with code) This guide demonstrates building an Oracle AWR-equivalent diagnostics stack in PostgreSQL using extensions like pg_profile, pg_wait_sampling, and pg_stat_statements, with runnable SQL for snapshot history, wait event sampling, Top SQL analysis, and buffer cache inspection.
- Top 10 Performance Optimization Techniques for Azure Database for PostgreSQL Flexible Server This article outlines ten effective methods such as choosing appropriate compute SKUs, enabling Query Store, utilizing PgBouncer connection pooling, consulting Azure troubleshooting resources, monitoring autovacuum, and optimizing storage and IOPS to boost the performance and reliability of Azure Database for PostgreSQL Flexible Server.
- SQL Server Deadlock: Build One With Your Own Hands This SQL Server article demonstrates how to manually create a deadlock between two transactions by updating competing tables in opposite orders.
- SQL Server Performance Office Hours Episode 71 In Episode 71, Erik Darling discusses travel preferences, conference locations, ClickUp overview, an Azure VM issue, and a table variables discussion during SQL Server Performance Office Hours.
- Faster, Safer Version Upgrades for Databases with Large Objects Azure Database for PostgreSQL now handles large-object metadata more efficiently during major version upgrades, reducing memory pressure and enabling successful upgrades of databases with up to 500 million large objects.
- Announcing Automatic Backup Immutability for Azure SQL Database and Azure SQL Managed Instance Azure SQL Database and Managed Instance now automatically protect up to seven days of recent point-in-time restore backups with immutability, defending against ransomware and accidental changes without configuration.
- T-SQL in SQL Server 2025: REGEXP_LIKE SQL Server 2025 adds REGEXP_LIKE and four related functions to provide native regular expression support for advanced pattern matching in T-SQL.
- HostNameInCertificate changes in Azure SQL Managed Instance affecting client connectivity Azure SQL Managed Instance will reduce SAN versions in instance certificates starting August 1, 2026, which may break client connections using VNet-local endpoints and the HostNameInCertificate parameter if not updated.
- MySQL vs PostgreSQL: spatial data types, functions, and migration pitfalls explained This guide explains how MySQL and PostgreSQL handle spatial data differently, with PostGIS offering more advanced GIS capabilities for location-aware applications while MySQL’s recent enhancements have narrowed the performance gap.
- Two Insert Exec Problems The video demonstrates how Insert Exec can cause blocking issues and prolonged lock contention on tables using a stored procedure that deletes and inserts data while waiting five seconds.
- Lessons Learned #551: Azure SQL Connection Timeouts: Three Things to Check Azure SQL connection timeouts may hide token acquisition or DNS resolution delays instead of indicating database problems; diagnose by capturing full error details and timing each connection phase.
- Nine Unusual Ways My Clients Use AI With SQL Server This article reveals nine unconventional ways real database teams leverage AI with SQL Server, including extracting business rules from legacy code and reverse engineering closed vendor schemas.
- AI-assisted Oracle-to-PostgreSQL schema conversion in Visual Studio Code The new AI-assisted schema conversion feature in Visual Studio Code’s PostgreSQL extension simplifies migrating Oracle databases to Azure Database for PostgreSQL by handling discovery, translation, validation, and review within a single project workflow, reducing manual effort and ensuring compatibility with complex Oracle constructs.
- Free SQL Server Performance Monitoring: Where Things Are And Where Things Are Going The free SQL Server Performance Monitoring tool now includes a storage growth tab that shows which database objects are expanding fastest, helping administrators right-size servers and optimize costs.
- Database Animations: Why Big Columns May Not Affect Logical Reads This article explains that adding many columns or wide data types increases row length without affecting single-row logical reads but may impact performance when reading multiple rows, shown through database animations comparing seek and scan methods.
- From 20+ seconds to 2 - faster SQL Database Project .dacpac deployments are finally here (preview) The new DacFx 170.5.60-preview release cuts SQL Database Project deployment time from over 20 seconds to about 2 seconds by caching model checksums and skipping redundant schema validation steps.
- How to ensure replication stability in PostgreSQL (fix/prevent lag, bloat, and bottlenecks) Ensuring replication stability in PostgreSQL involves tuning configuration settings, managing table bloat with regular VACUUM operations, and handling large DDL changes safely to prevent lag, network issues, and performance bottlenecks.
- mssql-django 1.8.0: Django 6.1 Support within 48 Hours of Django 6.1 GA mssql-django 1.8.0 rolled out Django 6.1 support within 48 hours of its general availability, enabling seamless upgrades for SQL Server, Azure SQL, and Fabric users with a single pip command.
- mssql-python 1.13.0: Arrow Bulk Copy, Smarter Tokens, Slimmer Wheels mssql-python 1.13.0 adds an Apache Arrow bulk copy feature streaming data directly to SQL servers with no Python overhead, supports Azure Identity credentials, and relocates ODBC driver binaries to a separate package for smaller files and easier updates.
🛠️ Developer tools
- Deduplicate Azure Bicep Parameter Files with Extendable Parameters Azure Bicep's extendable parameters let you define shared values in a base file and override only needed settings in environment-specific files, reducing duplication and configuration drift.
- Declarative Testing Helper for Marten or Polecat Projections The new EventProjectionScenario API in Marten and Polecat offers a declarative method for testing event sourcing projections with support for asynchronous projections and detailed control over data cleanup and execution timing.
- How to Use GitHub Copilot with Local LLMs in VS Code This guide compares GitHub Copilot AI credits, Azure AI Foundry API costs, and running local LLMs in VS Code to help developers select the most cost-effective option.
- Trying Out WSL Containers: Microsoft's Built-In Alternative to Docker Desktop WSL Containers provides a built-in method to run Linux containers on Windows with Docker-CLI compatibility, though its current limitations-slow performance for Windows file system repos and lack of support for Docker Compose and .NET Aspire-mean it is promising but not yet ready to fully replace Docker Desktop.
- How the GitHub legal team used Copilot CLI to streamline their workflows GitHub's legal team used the Copilot CLI to develop custom tools such as an internal drafting style guide and automated workflows, enhancing efficiency and consistency with minimal coding expertise.
- Running Visual Studio 2008 in Windows 11 Running Visual Studio 2008 on Windows 11 is possible by disabling Smart App Control, installing the trial from Microsoft, and enabling SP1 patches, allowing legacy projects to stay responsive while moving toward modern databases like SQLite.
- VS Code 1.132 Sharpens Agent Workflows with Side Chats, Browser Comments VS Code 1.132 introduces side chat and browser comment features to enhance agent workflows alongside other editor improvements.
- MCP safety & evaluation with the Agent 365 CLI & Agent Governance Toolkit The Agent 365 CLI assesses MCP servers for production readiness, scoring tool definitions, and offers a prioritized list of improvements; the Agent Governance Toolkit regulates sensitive actions during runtime to ensure safety and compliance.
- Native AOT in Uno Platform: faster startup on five platforms Uno Platform's Native AOT feature cuts app startup times by up to 60% on Android, iOS, Linux, macOS, and Windows through ahead-of-time compilation without needing architectural changes.
- TimescaleDB Support within Marten Marten now integrates TimescaleDB support, enabling automatic conversion of tables into hypertables with time-partitioned storage, columnar compression, retention policies, and continuous aggregates via a single UseTimescaleDB() call.
- A guide to slash commands in the GitHub Copilot app Slash commands in the GitHub Copilot app let you quickly manage sessions, navigate projects, and customize workflows, enhancing productivity by providing visual shortcuts unavailable in the CLI.
- Get started faster in the PostgreSQL extension for VS Code The new Get Started landing page in the PostgreSQL VS Code extension guides users through key features, resources, and quick actions after connecting to a server, enhancing productivity.
- VS Code Agent Plugins Go Cross-Client with New Open Standard VS Code now supports cross-client agent plugins through an open standard, enabling portable skills and multi-cloud management platforms to work across different IDEs.
- Writing Effective Copilot Instructions for Complex Codebases GitHub Copilot needs a modular skill-based approach with custom instructions for large, multi-project codebases like Corvus.Text.Json to efficiently access domain-specific knowledge across thousands of lines.
- Copilot Credit Complaints Keep Coming: 'Too Expensive to Use' Developers express frustration over rapidly vanishing AI credits under GitHub Copilot’s new usage-based billing model, raising concerns about affordability compared to alternative coding assistants.
- ChatClient Middleware vs. Agent Middleware ChatClient Middleware intercepts and logs messages between an agent and a language model, while Agent Middleware handles session tracking and token budgeting to optimize agent performance in the framework.
🔩 DevOps
- Turn one giant AI-generated pull request to a reviewable stack GitHub's new stacked pull requests break down large AI-generated code into smaller reviewable layers, enhancing collaboration and reducing merge conflicts for complex features.
- Azure DevOps Remote MCP Server is generally available The Azure DevOps Remote MCP Server provides secure, context-aware AI assistant access to Azure DevOps projects through a hosted endpoint, removing local installation needs and facilitating real-time integration with tools such as Visual Studio Code, GitHub Copilot, and Microsoft Foundry.
- You can now use the Azure DevOps Service Connection instead of a PAT or Build Session token The new Azure DevOps service connection uses secure workload-identity authentication via service principals or managed identities, improving security and simplifying secret management while allowing cross-organization access within the same Entra ID tenant.
- Pull request annotations on work item cards Azure Boards now integrates pull request annotations directly onto Kanban work item cards allowing developers to monitor code review status instantly within the board.
- How we took malware advisories beyond npm GitHub's Dependabot now flags malware across eight package ecosystems-including PyPI-by integrating OpenSSF's malicious-packages data into the GitHub Advisory Database.
- Missing Azure DevOps Server Team Project Collection database when doing an upgrade Users may lose team project collection database artifacts during Azure DevOps Server 2019 to Azure DevOps Services upgrade if they don't validate and clean up the source environment before uploading the backup.
- Announcing Grafana 13 Support in Azure Managed Grafana Azure Managed Grafana now supports Grafana version 13, adding features such as Dynamic Dashboards, Saved Queries, improved filtering, template support, and Git Sync for code-managed dashboards.
🧬 Hybrid + multicloud
- Public Preview: Azure ExpressRoute resiliency guard Azure ExpressRoute resiliency guard’s public preview introduces a new model property to configure gateways as single-homed or multi-homed, enhancing network reliability and redundancy.
- Unifying Local Credential Security – Legacy LAPS, Windows LAPS, and Azure Arc in Practice This guide unifies Local Administrator Password Solution (LAPS) implementations across Legacy LAPS, Windows LAPS, and Azure Arc, detailing a five-stage modernization process that reduces lateral movement risk by integrating hybrid and multicloud credential rotation.
- Build and Deploy Confidential Virtual Machines on Azure Local (Public Preview) Confidential Virtual Machines on Azure Local provide hardware-based memory encryption and attestation for secure execution of sensitive workloads on customer-owned edge infrastructure.
- Generally Available: Windows Server 2016 Extended Security Updates enabled by Azure Arc Azure Arc provides flexible pay-as-you-go Extended Security Updates for Windows Server 2016 with integrated asset inventory and seamless connection to Azure Update Manager, Change Tracking, and Policy for improved management and compliance.
- How to Build Your Own Azure Local Lab: From Laptop to Dedicated Server Choosing a Dedicated Lab Platform is crucial for Azure Local labs, as memory capacity-rather than CPU speed or expandability-determines how much you can run, with recommendations starting at 128 GB RAM for full functionality.
🎭 Identity
- Updated ECMA2Host Tools with Entra Application Provisioning Job Management The updated v1.2.0 of ECMA2HostTools now includes capabilities for managing Entra application provisioning jobs, addressing the need for better visibility and control over connector operations.
- How to prevent users from registering applications in Microsoft Entra Learn how to restrict application registration in Microsoft Entra for standard users, along with the security risks of disabling this feature.
- The memberOf rule operator is ending for dynamic Groups in Entra Microsoft will retire the Entra memberOf rule operator by October 2026, so administrators should use Graph PowerShell to identify impacted dynamic groups and plan an alternative solution in advance.
- End VPN gaps with identity-first access Zero Trust replaces traditional VPNs by continuously evaluating identity, device posture, location, and risk signals to ensure dynamic, least-privilege access across various resources without legacy architecture complexity.
- System-Preferred Authentication and September 1st Passkey Change Microsoft is set to switch default authentication from SMS to Passkeys on September 1, with a phased rollout allowing organizations to prepare and minimize user disruption.
- Entra ID Protection – User Risk with Unified Identity Signals from Defender is here! Entra ID Protection now integrates Defender signals to automate risk detection across endpoints and cloud identities for faster threat remediation.
- MFA vs. Conditional Access in Microsoft Entra ID MFA vs.
- Entra PIM vs. Delinea vs. CyberArk: Don’t Buy the Wrong PAM Entra PIM specializes in role management, whereas Delinea and CyberArk focus on credential vaulting, differing in session recording, audit evidence, and total costs over three years.
- Microsoft Entra Just Made Passwordless MFA Registration Easier Users can now register passwordless credentials and later choose device-bound MFA sign-in to streamline authentication.
🔌 Integration
- AsyncAPI Code Generation with Corvus: Custom Transports Custom transports for AsyncAPI allow integration with non-standard messaging systems such as Redis Streams, Amazon SQS, and Google Cloud Pub/Sub through a straightforward interface managing publishing, subscribing, request-reply, and dead-letter handling.
- Use connectors with Managed Identity in the Logic Apps Standard extension The Logic Apps Standard VS Code extension now supports Managed Identity authentication for connectors during local development and deployment, streamlining security and easing environment transitions without code changes.
- Update WCF Relay applications to use TLS 1.2 or later Updating WCF Relay applications to use TLS 1.2 or later is necessary because Azure has retired TLS 1.0 and 1.1, causing connectivity failures for listeners that still negotiate these older protocols.
💡 Internet of Things
- RTSP stream monitoring on the Edge and beyond via Azure IoT Operations Azure IoT Operations enables real-time monitoring of IP camera RTSP streams as JPEG images on edge devices, extending industrial video analytics capabilities beyond traditional data centers.
- City of Raleigh from insights to action: A traffic digital twin, live in six weeks Raleigh developed a real-time digital twin of its traffic using existing cameras, AI analytics, and Azure infrastructure to optimize signal timing and potentially save commuters $9.7 million annually without altering street assets.
- Turning Microsoft Fabric into a Factory Operations Control Center: RTSP snapshots in Fabric Apps This article shows how IP camera snapshots from factory edges are fed into Microsoft Fabric Real-Time Intelligence through Eventhouse, allowing operators to annotate alarms and build an integrated operations control center.
- Marine Protected Areas in Microsoft Fabric Maps Fabric Maps leverages real-time IoT data to overlay marine protected areas in the Mediterranean Sea, integrating predictive insights on sea mammal protection alongside remote monitoring capabilities.
🎓 Learning and Certifications
- Build Azure Skills Faster with Cloud Academy, a Visual Studio Subscriber Benefit Cloud Academy provides Visual Studio subscribers with flexible, hands-on Azure learning paths featuring certification-focused labs, interactive assessments, and AI guidance to enhance developer skills swiftly for both daily tasks and exam prep.
- Recent and upcoming Microsoft exam changes – Friday 7 August 2026 The article details recent and upcoming modifications to several Microsoft certification exams, including updates to the AZ-400 and SC-100 exam schedules as of July 27 and 28, 2026.
⚖️ Management and Governance
- Built-In Observability for Serverless AI Agents on Azure Functions Azure Functions now includes built-in OpenTelemetry tracing that automatically records every serverless AI agent run, including the model used, session details, and all tool calls or sub-agent interactions, eliminating the need for any custom telemetry code.
- From Alert to Resolved: Building a Self-Healing Azure Platform with SRE Agent Azure SRE Agent streamlines alert-to-resolution by integrating telemetry, correlating data, proposing fixes within permissions, and providing instant root cause analysis, reducing MTTR from hours to minutes.
- Zonal Resiliency in Azure: Application-Centric Goals, Recovery Plans, and Drills Resiliency in Azure offers a unified view of zonal performance with automated recommendations and recovery plans to minimize downtime and boost app reliability.
- Bicep Tips and Tricks | #11 | Working with Role Assignments Bicep's #11 article discusses common challenges in role assignments for Azure such as deterministic naming issues, GUID-based roleDefinitionIds, and enhancing deployment error messages to improve automation.
- Zero Ops: Agents Operate, Humans Govern Zero Ops moves operations from human system management to AI agents performing tasks with humans setting intent, overseeing outcomes, and verifying results via advanced models and strong governance.
🚌 Migration
- Agentic Migrations and Modernization: How the Azure Migrate Agent Keeps Your Intent Alive End to End The new Azure Migrate agent uses AI to preserve migration intent and context, automating guidance and reducing decision fatigue in large-scale cloud migrations.
🌐 Networking
- Generally Available: Azure Virtual Network routing appliance The Azure Virtual Network routing appliance offers private inter-VNet connectivity with hardware-accelerated low latency and high throughput, outperforming VM-based solutions.
- Generally Available: Explicit proxy in Azure Firewall Azure Firewall's explicit proxy feature enables direct HTTP/HTTPS traffic routing through the firewall using standard proxy settings, improving security flexibility for enterprise workloads.
- Public Preview: Azure DNS enables DNS-based load balancing through Traffic Manager integration Azure DNS’s public preview introduces direct integration with Traffic Manager, allowing users to link DNS records seamlessly and avoid the need for additional CNAME mappings.
- Public Preview: Azure Private Link support over IPv6 Azure introduces public preview of Private Link support over IPv6, allowing secure and private access to PaaS services like Azure Storage and SQL Database from IPv6-enabled virtual networks.
- Designing Azure Networks That Scale: From Small Deployments to Enterprise-Grade Designing Azure Networks That Scale focuses on planning for future growth using tools like Azure Virtual Network Manager to manage expanding VNets across teams and regions before manual efforts become unsustainable.
- Many-regions networking with Azure Route Server Azure Route Server optimizes multi-region networking by enabling dynamic route injection from NVAs in existing hubs without extra deployments, cutting costs and reliance on a single region.
- Azure Virtual Network routing appliance is now generally available Azure Virtual Network's routing appliance is now generally available, offering managed, platform-native routing at cloud scale with support for high-performance AI workloads, scalable hub-and-spoke architectures, extended private connectivity, and seamless IPv6 and multi-region deployments.
- 354 - A look at the new Azure Front Door edge actions Azure Front Door adds programmable edge actions enabling custom JavaScript for tasks such as A/B testing, token validation, and origin routing with sub-second latency.
- Announcing Public Preview - Azure Private Link over IPv6 Azure announces a public preview of Private Link over IPv6, enabling secure and efficient access to Azure PaaS services from IPv6-based on-premises environments via ExpressRoute connections.
🔐 Security
- Microsoft Defender for Cloud Customer Newsletter The latest Microsoft Defender for Cloud newsletter announces the GA of various container security features such as misconfiguration recommendations for Kubernetes, AKS upgrades, runtime vulnerability assessments for EKS and GKE, and improved scanning for hardened Docker containers.
- 128 Seconds to disruption: Microsoft Defender stops ransomware at QNET Microsoft Defender's new device isolation feature halted a ransomware attack at QNET in 128 seconds by quickly isolating compromised endpoints.
- Advance Zero Trust for AI: New tools and guidance to secure AI agents and DevSecOps Microsoft enhances its Zero Trust for AI strategy with an updated assessment tool and a new DevSecOps pillar to secure AI agents and workflows.
- ChainDrop supply chain compromise: Anatomy of a self-propagating worm ChainDrop exposes a sophisticated self-propagating worm using obfuscated Bun JavaScript payloads to steal credentials from developer workstations and CI/CD pipelines for authentication across npm, GitHub, AWS, Kubernetes, and HashiCorp Vault, then auto-republishs compromised packages.
- Azure Key Vault Premium and AES-KW Post Quantum Resistant Key Wrapping in Public Preview Azure Key Vault introduces AES-KW for quantum-resistant key wrapping in public preview, allowing services to upgrade from RSA-based management.
- Containment Strategies for AI now available in Secure Now Secure Now introduces containment strategies for AI agents to limit actions, harden surfaces, and improve oversight amid recent autonomous AI security incidents.
- Building a CCF Nested API Pull Connector: A Technical Lab Walkthrough This guide shows how Azure's CCF Nested API Pull Connector simplifies ingesting data from two-step enterprise APIs into Sentinel using a JSON config to automate list and detail endpoint calls, removing the need for custom orchestration code.
- Public Preview: Nested API Support Comes to Microsoft Sentinel CCF Nested API support in Microsoft Sentinel’s Codeless Connector Framework allows ISVs to integrate multi-step data retrieval seamlessly without modifying endpoints, enhancing connector coverage for list-then-detail API designs.
- Microsoft named a Leader in the KuppingerCole Leadership Compass for Cloud Native Application Protection Platforms (CNAPP) Microsoft earns Leadership status in KuppingerCole’s CNAPP report for its Defender for Cloud, which unifies cloud and AI security into a single platform supported by an advanced agentic AI ecosystem.
- Blogpost: Microsoft Security Copilot: A Revolution for Enterprises and SOC Analytics Microsoft Security Copilot boosts analyst productivity by 22% while integrating with existing tools for enhanced SOC analytics via real-time threat detection and automated responses.
- AI-powered playbook generator, now available to more customers The AI-powered playbook generator is now available to all Microsoft Sentinel customers for free, automatically creating editable playbooks from natural language descriptions including tests and documentation.
- HOWTO: Implement FBI CJIS 6.0-auditable configurations of Azure Commercial cloud The guide explains how agencies can set up Azure Commercial to comply with FBI CJIS 6.0 audit standards using Customer Managed Keys and advanced security, detailing a six-phase process from initial setup to ongoing monitoring and endpoint protection.
- Expanding the Azure Local Security Ecosystem with IBM QRadar and Splunk The integration of IBM QRadar and Splunk with Azure Local boosts security operations by incorporating Azure Local's signals into trusted monitoring platforms for regulated environments while maintaining existing workflows.
- Smarter signals, broader coverage: UEBA anomalies on top of Behaviors layer and new data sources Microsoft Sentinel's Behaviors layer now integrates UEBA anomaly intelligence and supports additional data sources like Fortinet, network, and cloud logs, enhancing analyst understanding and detection of suspicious activity.
📦 Storage
- Public Preview: Migrate from AWS FSx for Windows File Server to Azure Files with Azure Storage Mover Azure Storage Mover's public preview allows agentless migrations from AWS FSx for Windows File Server to Azure Files, streamlining the process by removing the need for separate migration tools.
🔗 Web
- Announcing public preview: Markdown for Agents in Azure App Service Markdown for Agents in Azure App Service reduces HTML noise by converting web content to text-focused Markdown, shrinking response size by 97 % on average while cutting conversion time under 2 ms, now available in public preview for Windows apps across all regions.
- Modern authentication in ASP.NET Core with 2FA and passkey This guide demonstrates how to integrate two-factor authentication (2FA) and passkeys into an ASP.NET Core API, enhancing security by requiring multiple verification factors beyond passwords.
- Automatic CSRF protection based on Fetch Metadata headers ASP.NET Core's new CSRF protection in .NET 11 preview 6 uses Fetch Metadata headers for automatic request verification, enhancing speed and security by minimizing token reliance.
- Build Durable, Long-Running MCP Tasks on Azure App Service Azure App Service now supports durable, long-running MCP Tasks using Azure Table Storage for task state and Azure Service Bus for executing work items, ensuring continuity through instance recycling or scaling.