This blog post series highlights the key announcements and major updates related to Azure Infrastructure as a Service (IaaS) and Azure Local, as officially released by Microsoft in the past two weeks.

Azure

General

Azure Copilot Troubleshooting Agent

Azure Copilot Troubleshooting Agent is now Generally Available (GA), providing a unified, built-in experience to help customers investigate and resolve operational issues more quickly. Available through both Azure Copilot and the Support + Troubleshooting experience in the Azure portal, the agent brings together troubleshooting insights, operational context, and recommended actions in a single workflow. This helps customers move more efficiently from issue detection to diagnosis and resolution, reducing the need to manually correlate information across multiple Azure tools and support experiences.

Microsoft named a Leader in the 2026 Gartner® Magic Quadrant™ for Container Management

Microsoft has been named a Leader in the 2026 Gartner® Magic Quadrant™ for Container Management and, according to Microsoft, is positioned furthest to the right for Completeness of Vision. Microsoft views this recognition as reflecting its strategy to provide a consistent container operating model across cloud, hybrid, edge, and sovereign environments while supporting both traditional applications and emerging Artificial Intelligence (AI) workloads. The company’s container portfolio spans Azure Kubernetes Service (AKS), Azure Container Apps, Azure Arc, and Azure Kubernetes Fleet Manager, enabling organizations to modernize existing applications, run AI workloads closer to data and users, and manage distributed Kubernetes environments without requiring applications to be redesigned for each deployment location.

Compute

Azure Ephemeral OS Disk with full caching for VM and VMSS

Azure Ephemeral OS Disk with full caching is now Generally Available (GA) for supported Azure Virtual Machines and Virtual Machine Scale Sets, providing higher performance and improved resiliency by caching the entire operating system disk image on local VM storage. Unlike the default Ephemeral OS Disk caching model, where some reads may still depend on remote storage, full caching serves all OS disk I/O locally once the background caching process has completed. This helps improve resilience during remote storage disruptions, provides consistently low-latency OS disk performance, and preserves VM provisioning speed because caching occurs asynchronously after startup.

The capability is particularly suited to I/O-sensitive stateless workloads such as AI training and inference, quorum-based databases, real-time analytics, and large-scale stateless services. Azure uses the VM’s local cache, resource disk, or NVMe storage to hold the cached OS image, reducing available local capacity by approximately two times the OS disk size plus 1 GB. Full caching is available across all Azure public regions for supported VM sizes with eight or more vCPUs, including N-, L-, M-, and H-series VMs, v5–v7 D-, DC-, E-, Eb-, and EC-series VMs, and v6–v7 F-series VMs. Microsoft states that there is no additional charge for the full-caching feature beyond the standard VM and applicable base OS disk costs.

Mdsv4 and Msv4 Series Virtual Machines for SAP (preview)

The Mdsv4 and Msv4 Series Virtual Machines are now available in Public Preview. These new memory-optimized VM families are built on 6th Generation Intel® Xeon® Scalable processors and incorporate advanced security capabilities together with the latest Azure Boost technologies. Designed for demanding memory-intensive and mission-critical workloads, including SAP environments, the new series are intended to provide improved performance, scalability, efficiency, and reliability for large-scale enterprise applications while helping customers optimize infrastructure costs.

Networking

High-scale mesh in Azure Virtual Network Manager

High-scale mesh using connected groups in Azure Virtual Network Manager is now Generally Available (GA). In supported regions, customers can connect up to 3,000 virtual networks within a single mesh connectivity configuration by default, with higher-scale IP connectivity available by request. The capability allows organizations to establish direct network-to-network connectivity across very large Azure environments without creating and maintaining individual virtual network peerings. High-scale mesh uses the existing network group and connectivity configuration model, making it easier to keep applications, environments, and business units logically separated across virtual networks while simplifying connectivity management at enterprise scale.

Network Planning in Azure Migrate (preview)

Azure Migrate Network Planning is entering Public Preview, extending Azure Migrate assessments and business cases with network-aware planning for VMware lift-and-shift migrations. The capability is designed to bring networking considerations into the same workflow already used for workload discovery, readiness assessment, migration planning, and cost estimation, helping customers understand not only which virtual machines can move to Azure, but also how application connectivity, security requirements, and network dependencies should be represented in the target environment.

Network Planning uses discovered VMware networking information to provide context around workload placement, connectivity, dependencies, and applicable Azure destination constructs. It translates source findings into Azure-native networking recommendations, including considerations for hub-and-spoke topology, mapping on-premises firewall intent to Network Security Groups (NSGs) and Azure Firewall, and application delivery requirements such as load balancing and Web Application Firewall (WAF) policies. Assessments can also surface network-related readiness issues that require design decisions or remediation before migration, helping teams distinguish straightforward workload moves from scenarios that need additional engineering.

Networking is also incorporated into the Azure Migrate business case, allowing estimated network costs to be evaluated alongside compute, storage, and other migration expenses. Where sufficient source information is available, security-rule context can also help application, network, security, and migration teams understand existing policy intent and prepare target-state controls. The objective is not to automatically design the target network, but to provide a structured, evidence-based starting point that reduces reliance on fragmented information from vCenter, NSX, firewalls, load balancers, spreadsheets, and operational knowledge.

The initial preview is focused on lift-and-shift migration planning, while capabilities such as application modernization, advanced network topologies, third-party virtual appliances, infrastructure-as-code generation, and IP-retention scenarios remain future roadmap areas. By adding network evidence and estimated costs directly into Azure Migrate, Microsoft aims to make migration planning more complete and reduce the risk of networking becoming a late-stage critical path during VMware-to-Azure migration projects.

HTTP/3 over QUIC support in Azure Application Gateway (preview)

Azure Application Gateway now supports HTTP/3 over QUIC in Public Preview, enabling frontend client connections to use the latest HTTP transport model for improved connection establishment, lower latency, and greater resiliency. Because QUIC supports faster handshakes and independent streams, it can reduce the impact of packet loss and improve responsiveness over unreliable or high-latency networks. This makes the capability particularly relevant for latency-sensitive web applications, messaging platforms, mobile experiences, transactional systems, and APIs. HTTP/3 is configured at the listener level, allowing customers to enable it selectively on Basic listeners, and during the preview it can be configured through the Azure portal or REST API.

Storage

User-bound user delegation SAS for Azure Storage

User-bound user delegation Shared Access Signature (SAS) for Azure Storage is now Generally Available (GA) in all public regions, introducing a more restrictive authorization model for delegated access. While standard user delegation SAS already binds token creation to an authenticated Microsoft Entra identity, user-bound user delegation SAS also restricts token usage to a specific end-user identity. This helps ensure that a delegated SAS token can only be used by its intended user, providing stronger protection than account SAS, service SAS, or standard user delegation SAS. The capability extends the existing user delegation model available across Azure Blob Storage, Azure Files, Azure Tables, and Azure Queues, while retaining features such as Entra-based token issuance, traceability to the delegator, and a maximum token validity of seven days. There is no additional charge beyond the standard transaction costs associated with the storage account.

Azure Local

Microsoft named a Leader in the 2026 Gartner® Magic Quadrant™ for Distributed Hybrid Infrastructure

Microsoft has been recognized as a Leader in the 2026 Gartner® Magic Quadrant™ for Distributed Hybrid Infrastructure (DHI) for the fourth consecutive year and is positioned highest for Ability to Execute. Microsoft views this recognition as reflecting its unified approach to distributed infrastructure, built primarily around Azure Local and Azure Arc, which provides a consistent Azure management, governance, and operational model across datacenters, edge locations, multicloud environments, and sovereign deployments. The platform supports different infrastructure architectures, including hyperconverged and disaggregated models, while allowing organizations to run workloads where business, regulatory, latency, or data-residency requirements dictate. As distributed infrastructure increasingly needs to support traditional applications alongside data and AI workloads, Microsoft continues to invest in connected and disconnected operations, scalable infrastructure, sovereign capabilities, and consistent cloud-to-edge management.

Azure Local workload management in Azure Arc Site Manager (preview)

Azure Local workload management in Azure Arc Site Manager is now available in Public Preview, providing a centralized experience to monitor and operate virtual machines and Kubernetes clusters across multiple Azure Local deployments. Azure Arc Site Manager organizes distributed infrastructure into Sites that reflect physical locations such as datacenters, factories, retail stores, and branch offices, allowing operators to manage workloads in the context of where they actually run rather than navigating across subscriptions, resource groups, and individual Azure Local instances.

The new workload management experience provides a unified view of Azure Local virtual machines and Kubernetes clusters, including operational status and key workload information, making it easier to identify unhealthy resources or investigate issues at a specific location. Operators can also perform bulk actions across multiple workloads: virtual machines support bulk start, restart, stop, pause, save, and delete operations, while Kubernetes clusters can also be deleted in bulk. This can significantly reduce operational effort for recurring tasks such as bringing a site back online after maintenance, shutting down workloads at scheduled times, or decommissioning resources from a retired location.

Site Manager also improves navigation across large distributed estates through grouping, filtering, and search capabilities. Administrators can group workloads by attributes such as the Azure Local instance on which they run and then filter or search by name to quickly identify specific resources. By combining centralized visibility, site-aware organization, and bulk workload operations, Azure Arc Site Manager provides a more scalable operational model for organizations managing Azure Local infrastructure across multiple edge and datacenter locations.

Azure Local: Features and improvements in 2609

Microsoft has released the September 2026 update for Azure Local, identified as version 12.2609.1003.7, with OS build 26100.33438. This release focuses primarily on reliability fixes and operational corrections affecting Azure Local virtual machines and management services.

For Azure Local VMs, Microsoft has fixed an issue affecting clusters earlier than version 2604 where the az stack-hci-vm stop command could fail when stopping a running virtual machine with the error “Cannot do StopGraceful power operation on a Running VM.” The command now remains compatible with pre-2604 clusters, allowing administrators to stop running VMs without encountering this error.

The release also addresses an Azure Machine Configuration billing issue that caused some Azure Local clusters to be incorrectly charged for the service, even though Azure Machine Configuration should be available at no additional cost for Azure Local customers. Microsoft notes that this issue was resolved through the August 2026 Azure Machine Configuration agent update.

Conclusion

Over the past two weeks, Microsoft has introduced a slew of updates and announcements pertaining to Azure Infrastructure as a Service (IaaS) and Azure Local. These developments underscore the tech giant’s unwavering commitment to enhancing its cloud offerings and adapting to the ever-evolving needs of businesses and developers. Users of Azure can anticipate improved functionalities, streamlined services, and enriched features as a result of these changes. Stay tuned for more insights as I continue to monitor and report on Azure’s progression in the cloud sphere.