Azure Weekly Update covers recovery and data reads
Principal Cloud Solutions Architect
John Savill’s Azure roundup covers VM file recovery, faster data-lake reads and identity telephony options, helping teams spot changes to assess.
Key insights
- Azure Backup item-level recovery for Premium SSDv2 and Ultra Disk VMs is reported as generally available.
- ABFS Vectored Read groups nearby file ranges and reads asynchronously to reduce storage-request overhead.
- Choose Your Own Telephony for Microsoft Entra ID is scheduled to start becoming available on October 30, 2026.
- The video chapter list also includes Azure Monitor, Cosmos DB, Traffic Manager and Copilot Studio updates.
John Savill’s [MVP] published an Azure Weekly Update video covering a selection of Azure and AI announcements. Shown on October 24, 2026, the 15-minute roundup spans backup, data access, networking, monitoring, identity and other topics. The practical takeaway is that several changes may be relevant to operations teams, but the video description provides chapter names rather than detailed explanations for most of them.
What's new
The research notes describe two Azure changes as generally available: item-level recovery for Azure virtual machines using Premium SSDv2 and Ultra Disks, and ABFS Vectored Read for Azure Data Lake Storage. Item-level recovery lets administrators retrieve individual files and folders instead of restoring an entire VM. The notes say it supports both 512-byte and 4-KB logical sector sizes.
ABFS Vectored Read is intended for compatible analytics engines that need selected parts of files. It combines nearby ranges into fewer storage requests and performs reads asynchronously. The aim is to reduce request and network-round-trip overhead for workloads that do not need to read whole files.
The roundup also includes Microsoft Entra Choose Your Own Telephony, or CYOT. The research notes say its availability is scheduled to start on October 30, 2026, and describe it as an option for choosing a telephony provider in Entra-related identity scenarios. The video chapters additionally name Traffic Manager and Azure DNS integration, Application Gateway keyless TLS with Azure Key Vault managed HSM, Azure Front Door WAF service-tag matching, Azure Monitor updates, Cosmos DB safe key rotation, and a Copilot Studio agent ID.
Why it matters
For infrastructure administrators, item-level recovery can make a restore more targeted when the problem affects a small set of files rather than a complete virtual machine. That distinction can avoid undertaking a full-VM restore for a limited recovery need. The available information does not quantify recovery-time improvements or describe the setup process, so teams should treat the benefit as a change in recovery granularity rather than a promised time saving.
Data engineers and analytics teams may want to check whether their engines and workloads can use ABFS Vectored Read. The mechanism is most relevant where queries retrieve selected ranges from files, because grouping nearby reads can reduce the number of requests. The notes do not name supported engines or explain whether application-side configuration is needed.
Identity administrators planning phone-based authentication should track the CYOT availability details and assess whether provider selection affects their environment. The material does not give the supported provider list or configuration requirements. Other chapter topics could matter to network, security, database and AI teams, but the supplied description alone does not establish what changed in those areas.
How it works
Item-level recovery changes the scope of a VM backup restore: instead of recovering the whole virtual machine, an administrator can recover selected files or folders. The stated disk coverage is Premium SSDv2 and Ultra Disks, with support for the two logical sector sizes noted above. No operational steps or prerequisites are included in the available material.
ABFS Vectored Read works by collecting nearby requested file ranges and retrieving them through fewer requests, with reads performed asynchronously. In practice, the described sequence is:
- An analytics engine requests selected ranges from a file.
- Nearby ranges are combined to reduce separate storage requests.
- The reads proceed asynchronously, reducing round trips for compatible workloads.
CYOT is presented as a choice of telephony provider for Entra-related scenarios. The available notes do not provide a configuration walkthrough, so administrators cannot infer the precise setup from this roundup alone.
Limitations and open questions
The video description lists chapters, but it does not include a transcript or a detailed summary of most announcements. As a result, the chapter names for networking, monitoring, Databricks, Cosmos DB, sovereign updates, Project Perception and Copilot Studio should be treated as pointers to topics, not as verified descriptions of specific feature behavior.
The supplied notes do not identify the compatibility requirements for ABFS Vectored Read, the operational prerequisites for item-level recovery, or the telephony providers and rollout scope for CYOT. They also do not provide licensing details. Readers should verify those details against the relevant Microsoft documentation before changing production configurations.
For more context on the roundup itself, see the Azure Weekly Update video. HubSite 365 also tracks other Azure Weekly Update coverage.
Sources
Frequently asked questions
Which VM disks are covered by item-level recovery?
The research notes identify Premium SSDv2 and Ultra Disks, with 512-byte and 4-KB logical sector sizes.
What does ABFS Vectored Read change?
It combines nearby file ranges into fewer storage requests and performs reads asynchronously. The material does not name compatible analytics engines.
When is Choose Your Own Telephony scheduled to become available?
The notes say availability is scheduled to start on October 30, 2026. They do not specify provider options or rollout scope.