> ## Documentation Index
> Fetch the complete documentation index at: https://help.nops.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Upgrade Azure VMs from Previous-Generation Series

> Move Azure VMs on older series (A, D v2, F, D v3, E v3 and others) to the same-size current v5 series.

Older Azure VM series such as A v2, D v2, DS v2, F, Fs v2, D v3 and E v3 cost more per vCPU than the current v5 series. nOps finds VMs on these older series running on pay-as-you-go and estimates the saving from resizing them to the matching current size (for example D v2 to Ds v5 or Es v5, A v2 to Dls v5, F to Fs v2, and D v3 or E v3 to Ds v5 or Es v5).

| | |
| - | - |
| **Cloud** | Microsoft Azure |
| **Category** | Compute |
| **Service** | Virtual Machines |
| **Action** | Modify |
| **Effort** | Medium |
| **Restart needed** | Yes |
| **Can be rolled back** | Yes |
| **Analysis window** | Last 7 days |
| **Source** | nOps |

## Why it matters

Older series are slower and dearer than the current generation, and they are retired over time. Moving early saves money and avoids a forced migration later.

## How nOps finds it

nOps reads the last 14 days of Azure VM and scale set billing. A VM is flagged when it ran on pay-as-you-go on at least 7 of those days, was billed in the last 2 days, and is not mostly covered by reservations. Its 7-day pay-as-you-go compute cost, scaled to a month, is multiplied by the difference between the old size's list price and the target size's list price in the same region and operating system. The target must be at least 3% cheaper, the service must not be mostly paid with credits, and the saving must be at least \$5 a month. For D v3 and E v3 VMs where the AMD size is as cheap or cheaper, nOps shows the AMD recommendation instead.

## Before you start

Permission to resize the VM, and a maintenance window: resizing restarts it.

## How to implement

1. Check the target size shown in the recommendation. Sizes without a local temp disk need the d variant if the VM keeps data on its temp drive.
2. Check the target size is offered in the VM's region and availability zone, and that it supports the VM's disk types and network settings.
3. Resize the VM (portal: **VM → Size**, or `az vm resize --resource-group RESOURCE_GROUP --name VM_NAME --size SIZE`). If the size is not offered, deallocate the VM first. For a scale set, change its size and update the instances.
4. Update Terraform, Bicep or ARM templates so a later deployment does not change it back.

## Risks and considerations

* Resizing restarts the VM. Data on the temporary disk is lost.
* Some older VMs use generation 1 images or settings that a newer size does not support; check before resizing.
* Reservations for the old size may not cover the new one.
* To roll back, resize back to the original size.

## FAQs

> [accordion] Why does a D v2 VM sometimes map to an E series size?
>
> D11 to D15 v2 are memory-optimized sizes. Their current equivalent is in the E series, which keeps the same memory per vCPU.

## Related

* [Move Azure VMs to AMD-based Series](https://help.nops.io/inform/supported-recommendations/move-azure-vms-to-amd-based-series)
* [Right-size Azure Virtual Machine](https://help.nops.io/inform/supported-recommendations/right-size-azure-virtual-machine)
* [Supported recommendations](https://help.nops.io/inform/supported-recommendations.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.