Prepare an Azure VM for retirement by identifying its exact size series and retirement date, selecting a supported replacement that fits the workload and region, testing the move outside production, and validating every dependency after the change. Do not wait for the deadline: Microsoft says VMs left in a retired series are deallocated, stop working, lose service-level agreement (SLA) coverage and support, and stop accruing charges.
The practical sequence is inventory, lifecycle verification, workload assessment, destination selection, quota and regional checks, nonproduction testing, a controlled production change, and post-change validation. The exact resize operation, downtime and prerequisites vary by VM family, operating system, disk layout and workload, so use the family-specific Microsoft notice alongside this general process.
As an Amazon Associate I earn from qualifying purchases.
Understand what Azure’s lifecycle labels mean
Azure VM sizes move through four lifecycle stages. The label attached to a size series determines whether you are planning normally, modernizing ahead of time, or working against a firm operational deadline. Microsoft defines the stages in its VM lifecycle overview.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →| Stage | Meaning for administrators | What to do |
|---|---|---|
| Current | The series is the recommended choice for new deployments. | Use it when it satisfies the workload’s technical and commercial requirements. |
| Extended | The series remains supported while customers evaluate newer options. | Plan a move to a Current series rather than starting new dependencies on the older family. |
| End of Life | Microsoft has announced retirement and supplied a retirement date. | Complete modernization before that date. Microsoft’s wording is to “Prepare and complete modernization plans before the retirement date.” |
| Retired | The series can no longer be created, used, or selected as a resize target. | Any remaining workload must already have been moved; a retired VM cannot continue running as a supported deployment. |
A capacity restriction is a separate condition, not proof that a series has been retired. A family can have limited capacity and also have a retirement announcement, or have a capacity constraint without a retirement date. Check the lifecycle notice and the capacity information independently.
#1 Best Overall
Start with an accurate VM inventory
You cannot make a reliable migration plan from a subscription-level guess. Build an inventory that ties each VM to an owner, a service and a business deadline. Include stopped VMs and machines managed through scale sets or automation, not just the instances visible in a production dashboard.
Record the fields that affect the decision
- Subscription, resource group, VM name and management owner.
- Exact VM size and size series, including whether the machine is an instance in a virtual machine scale set.
- Azure region, availability zone or region-pair design, and disaster-recovery location.
- Operating system image, attached disks, encryption configuration and accelerated-networking or other enabled capabilities.
- Workload role, service dependencies, traffic pattern, performance baseline and business criticality.
- Installed agents, VM extensions, backup and monitoring integrations, and any licensing or appliance requirement.
- Maintenance constraints, change owner, rollback contact and the latest date on which production work can safely occur.
Export the inventory to a versioned record rather than relying on a one-time portal view. Record the date and source of the lifecycle decision so a later operator can tell which announcement was applied to each VM.
Verify the applicable retirement notice and date
Compare every recorded size series with Microsoft’s live retirement overview and then open the family-specific announcement. Confirm that the notice applies to your exact series, region and deployment type. Read any replacement guidance, exceptions and links to migration instructions before selecting a target.
Check for changed plans
Retirement plans and regional availability can change. Recheck the official notice when you open the change request, immediately before testing, and again before the production window. Keep the announcement’s effective date in the ticket, but do not treat an old copy as permanently authoritative.
Use the date as an operational deadline
Schedule completion early enough to allow a failed test, quota request or replacement-family change. A resize attempted on the deadline day leaves no practical time to solve a regional-capacity or compatibility problem. If a VM reaches the Retired stage, Azure deallocates it; it stops serving workload traffic, no longer incurs VM charges, and no longer has the SLA or support that applied before retirement.
Rank #2
Separate retirement from capacity constraints
Capacity and lifecycle are different questions:
- Lifecycle question: Is Microsoft ending support or availability for this size series on a stated date?
- Capacity question: Can Azure allocate the requested size in the selected region and zone right now?
A Current destination can still fail allocation because a region or zone lacks capacity. Conversely, a series may have available capacity while approaching retirement. Treat both checks as gates in the plan, and document the result for each VM rather than assuming one answer covers the other.
Choose a supported destination around the workload
There is no universal one-size replacement. Select a Current series where possible, or an Extended series only as an interim choice while a modernization plan is underway. The destination must satisfy the workload, not merely have a similar name or vCPU count.
Compare the dimensions that can change behavior
| Decision area | Questions to answer before approval |
|---|---|
| Compute and memory | Does the target provide enough cores, memory, processor features and sustained throughput for peak and background workloads? |
| Storage | Are the required disk types, throughput, IOPS limits, caching modes and temporary-disk behavior supported? |
| Networking | Does the target support the required virtual network features, accelerated networking, IP configuration, bandwidth and load-balancer behavior? |
| Extensions and agents | Will backup, monitoring, security, configuration-management and custom script extensions install and report correctly? |
| Availability design | Is the size offered in the required region and availability zone, and can the deployment preserve the intended zone and failover topology? |
| Quota | Does the subscription have enough regional vCPU and family quota for the target, including surge capacity during a parallel migration? |
| Billing | What is the target’s published pricing and how do licensing, disks, reserved capacity or temporary duplicate instances affect the change cost? |
Microsoft’s retirement guidance recommends checking workload requirements, regional availability and subscription quota before migration. Exact performance and cost differences are workload-specific; do not infer them from the family name alone.
Plan for regional and zone availability
Check the target size in the actual region and, where relevant, the exact availability zone. A replacement listed in a general family table may not be deployable in your region. If the target is absent, evaluate another supported family, a different zone or a temporary region strategy only after reviewing data-residency, latency and disaster-recovery requirements. Microsoft’s 2026 retirement announcement warns that suggested replacements may not be available in every region.
Confirm quota before you schedule migration
Quota is a subscription limit, not a promise that capacity exists. Check regional compute quota for the destination family and leave headroom for the existing VM while a parallel copy, staging VM or scale-set rolling upgrade is active. If quota is insufficient, request an increase early enough for approval and propagation, then verify the new limit in the same subscription and region.
Rank #3
Include quota for failover and rollback. A plan that has just enough capacity to create the replacement may leave no room to keep the original online, restore a failed instance, or handle a scale-out event during the change window.
Assess technical compatibility before touching production
Build a short compatibility checklist from the workload owner’s requirements and the target family’s documentation. The Microsoft Infrastructure lifecycle FAQ recommends validating changes in test environments, while retirement guidance specifically calls out performance, networking, extensions and availability-zone support.
Performance
- Capture representative CPU, memory, disk latency, IOPS, throughput, network throughput and application latency at normal and peak load.
- Identify processor-specific behavior, licensing limits, NUMA assumptions, device pass-through or specialized instructions that the target must provide.
- Define pass/fail thresholds before testing, such as maximum request latency, batch completion time or error rate.
Networking
- Verify virtual network, subnet, network security group, route-table and private-endpoint behavior.
- Check accelerated networking and any dependency on a particular network interface driver.
- Test load balancers, health probes, public IP associations, DNS registration and east-west traffic between tiers.
Extensions, agents and operations
- Confirm that VM extensions, backup agents, monitoring agents, endpoint protection and configuration tools support the target operating system and architecture.
- Check boot diagnostics, guest health reporting, patching and certificate or secret retrieval.
- Verify that automation identifies the new size or resource ID correctly rather than assuming the old family.
Availability and recovery
- Confirm that the destination supports the required zone and that the VM, disks, load balancer and dependent services remain aligned.
- Run a restore or failover rehearsal if the migration changes the recovery design.
- Document the rollback point, data-consistency method and person authorized to stop the change.
Test the resize or migration outside production
Use a representative nonproduction VM or a controlled clone. The test should exercise the same image, extensions, disks, network path and application configuration as production. A trivial test VM can prove that Azure accepts a size, but it cannot prove that the real workload will perform or that its agents will remain healthy.
- Freeze the test baseline. Record the source size, image version, disk configuration, extension versions, network settings, zone and measured workload results.
- Prepare the destination. Confirm regional capacity and quota, then create or resize the test VM using the proposed target and the same security and monitoring controls.
- Run functional checks. Boot the guest, verify disks and services, exercise application transactions, and inspect extension and agent status.
- Run load and failure checks. Compare resource usage and latency with the baseline, test expected peak behavior, and verify health probes, failover and recovery paths.
- Record exceptions. Capture driver changes, unsupported extensions, altered device names, performance regressions and any required configuration updates.
- Approve or revise. Proceed only when the workload owner accepts the measured results and the rollback procedure has been rehearsed.
Do not generalize test results from one VM to every machine in the estate. Different disk counts, zones, images and traffic patterns can expose different constraints.
Plan the production change and rollback
Choose a change window that matches the workload’s maintenance agreement and the specific family guidance. Some operations require a stop and deallocation; others can be performed through a rolling strategy or by bringing up a replacement in parallel. Do not promise a universal downtime duration.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallBefore the window
- Take and verify the backups or snapshots required by your recovery policy.
- Confirm the target’s availability, quota and current retirement notice again.
- Notify application, network, security and support owners, and define who can authorize rollback.
- Pause conflicting automation such as autoscale, patching or deployment pipelines.
- Prepare exact commands or portal steps, validation queries and escalation contacts in the change record.
During the change
- Drain traffic or stop application writes according to the service’s consistency procedure.
- Perform the documented resize, replacement or rolling migration for the specific VM family.
- Watch provisioning, boot diagnostics, guest health, extension status and platform activity logs.
- Keep the original recoverable until validation passes and the rollback period expires.
Rollback triggers
Define objective triggers before starting: failed boot, missing disk, extension failure, networking loss, health-probe failure, data inconsistency, or a measured performance threshold outside the approved range. If a trigger occurs, stop rollout, restore traffic to the known-good path where safe, preserve logs, and reassess the destination rather than repeatedly retrying an unverified change.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Validate the result before declaring success
Validation must cover the platform and the application. A VM that is running but unreachable, missing an agent or slower than the agreed threshold is not a completed migration.
- Compute and guest: Confirm the reported size, boot completion, operating-system health, time synchronization and expected services.
- Storage: Check every data and temporary disk, mount point, filesystem, caching mode, encryption state, throughput and application-level read/write tests.
- Networking: Test inbound and outbound paths, name resolution, private links, security rules, load-balancer probes and inter-tier connectivity.
- Extensions and monitoring: Verify backup, monitoring, security and configuration agents are healthy and reporting to the intended workspace or service.
- Performance: Compare agreed CPU, memory, latency, IOPS, throughput and error-rate measurements with the pre-change baseline.
- Availability zones: Confirm the VM and attached resources are in the intended zone and that failover behavior still meets the design.
- Application: Have the service owner run user journeys, batch jobs, scheduled tasks and integration checks.
Keep the old resource, snapshot or deployment artifact only for the period permitted by your retention and security policies. Then remove obsolete resources, update diagrams and runbooks, close monitoring exceptions, and record the new size, region, zone, owner and validation evidence.
What happens if you miss the retirement date?
Microsoft states that VMs remaining in a retired series are deallocated and stop working. They stop incurring charges for the VM compute resource, but that is not a benefit: the workload is unavailable and the retired configuration has no SLA or support. Disks, snapshots, public IPs and other resources may have their own billing and lifecycle, so review them separately rather than assuming every associated resource disappears.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteAfter retirement, you cannot create a new VM in that series or resize another VM into it. Recovery therefore depends on preparation completed before the date, such as a tested replacement, a restorable image or a documented redeployment path.
Best Value
Current examples and how to interpret them
Microsoft’s 2026 VM size retirement announcement, accessed October 13, 2026, lists NP-series, HC-series and HBv2-series retirements effective May 31, 2027, with suggested replacement families. It also notes that plans can change and that replacements may not be available in all regions. Use those entries as examples, not as a complete permanent list; verify your own family and region on the live page.
The lifecycle overview gives Dv3, Dsv3, Ev3 and Esv3 as an example of End of Life series with a November 15, 2029 retirement date. The same page says capacity-growth restrictions for those examples began in July 2026. Those dates describe the cited examples and must be rechecked before a change is scheduled.
For a family-specific procedure, consult the applicable announcement such as Microsoft’s DCsv2-series retirement guidance, then apply the inventory, testing and validation controls described above.
Free tools Windows power users keep installed
One-click scans. No signup required.
A repeatable readiness checklist
- Every VM and scale-set instance has an owner, size series, region, zone, workload and criticality recorded.
- The exact lifecycle stage, retirement date and latest official notice are attached to the change record.
- Capacity restrictions were checked separately from retirement status.
- A Current or appropriately temporary Extended destination was selected using workload requirements.
- Regional availability, availability-zone behavior and subscription quota were confirmed.
- Performance, storage, networking, extensions, monitoring and recovery requirements were tested outside production.
- The production window, communication plan, backups, rollback triggers and authorization path are documented.
- Post-change checks cover platform health, application behavior, performance, networking, extensions and zone placement.
- Operational records, monitoring, runbooks and retirement evidence are updated after acceptance.
Following this sequence turns an Azure retirement notice into a controlled modernization project instead of an emergency outage. Recheck Microsoft’s lifecycle pages before each major decision because dates, supported families and regional availability can change.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
