Welcome to the world of Azure Local!
Azure Local (the artist formerly known as “Azure Stack HCI”) lets you run VMs, containers and a handful of PaaS services on-prem while Azure Arc keeps everything in the same control plane. Think of it as cloud super-powers wrapped in a 2U pizza box – minus the pineapple. (learn.microsoft.com)
1. What You’ll Need Before You Even Think About Powering On
Area | Key checklist items |
---|---|
Hardware | Vendor-validated nodes (1 – 16), TPM 2.0, 32 GB ECC RAM min, 200 GB OS drive, ≥ 1 Gb mgmt NIC + 2 × 10/25/40 Gb for data (learn.microsoft.com) |
Network | Management VLAN, (optionally) storage VLANs, routable DNS, outbound 443 to Azure, accurate NTP |
Active Directory | New OU, unique hostnames, domain functional level 2016+ (learn.microsoft.com) |
Azure side | Subscription owner, resource group, registered resource providers, RBAC roles (Azure Connected Machine Onboarding , etc.) (learn.microsoft.com) |
Time | ~ 2 hrs for a single node, ~ 2.5 hrs for a 2-node cluster (excluding coffee) (learn.microsoft.com) |
Pro tip: Do the prerequisites once in a build runbook; future clusters become copy-paste easy.
2. Stage 0 – Lay Down the Azure Local OS
- Download the ISO from the Azure portal (free 60-day trial if you need it).
- Boot each node and run the Install Azure Stack HCI wizard → Custom install → pick the blank drive.
- First boot asks for a new local admin password (12+ chars, upper/lower/number/special).
- Reboot happens twice – totally normal.
- Install vendor firmware/drivers; drop the Solution Builder Extension in
C:\SBE
if your OEM ships one. (learn.microsoft.com)
3. Stage 1 – Baseline Config with SConfig
Immediately after login, SConfig pops up automatically:
# Handy menu numbers
1) Domain/Workgroup 8) Network Settings
7) Remote Desktop 12) Reboot
- Configure each NIC with the right VLAN/IP, gateway and DNS that points to your AD.
- Set a valid NTP server:
w32tm /config /manualpeerlist:"pool.ntp.org" /syncfromflags:manual /update
- Rename the computer if you like, but do NOT join the domain yet. (learn.microsoft.com)
4. Stage 2 – Register Nodes with Azure Arc
On every node (run as local admin):
# 1. Log in & set context
Connect-AzAccount -SubscriptionId "<subId>" -TenantId "<tenantId>" -DeviceCode
$token = (Get-AzAccessToken).Token
$id = (Get-AzContext).Account.Id
# 2. Kick off Arc bootstrap
Invoke-AzStackHciArcInitialization `
-SubscriptionID "<subId>" `
-ResourceGroup "rg-azlocal" `
-TenantID "<tenantId>" `
-Region "westeurope" `
-ArmAccessToken $token `
-AccountID $id
When the script finishes, each server shows up in Azure as Machine – Azure Arc. (learn.microsoft.com)
5. Stage 3 – Run the Azure Local Deployment Wizard
In the Azure portal → Azure Arc | Azure Local → Get started → Deploy:
- Select machines you just registered → Install extensions → Validate.
- Key Vault – create or reuse; secrets & BitLocker keys live here.
- Networking – choose an Intent template (e.g., “Mgmt + Compute” on two 25 Gb NICs, “Storage” on two RDMA NICs).
- Management – point to your AD domain, OU, and supply deployment + identical local admin creds.
- Security – pick Recommended unless you enjoy sleepless nights.
- Volumes – let the wizard create one infra + one thin workload volume per node (you can fine-tune later).
- Validation → Review + create → Create.
Grab another coffee; single-node finishes in ~ 90 min, two-node in ~ 150 min. (learn.microsoft.com)
6. Stage 4 – Post-Deployment TLC
Task | Why / How |
---|---|
Verify Resources | Resource Group should show: 1 × Azure Local, 1 × Arc Resource Bridge, Key Vault, Custom Location, Storage accounts, plus one Machine resource per node. (learn.microsoft.com) |
Enable Health Alerts | Turn on capacity alerts when Storage Pool hits 70 %. |
Enable RDP (if needed) | Enter-PSSession <node>; Enable-ASRemoteDesktop – disable again when done. (learn.microsoft.com) |
RBAC configuration | Assign Azure Stack HCI Administrator / VM Contributor etc. to your ops team. (learn.microsoft.com) |
Create extra volumes / VM images | Use Arc-enabled VM or AKS-H commands once infra is green. |
7. Keeping It Fresh – Updates & Upgrades
- Monthly cumulative & semi-annual feature updates land via the Azure Update Manager; do not use SConfig, WAC or cluster-aware updating for the OS anymore. (learn.microsoft.com)
- Stay within 6 months of the latest feature build; October 31 2025 is end-of-support for 23H2.
- The orchestrator patches OS, Arc agents and (when your OEM participates) firmware in one go
8. Troubleshooting & “Oops” Recovery
- Deployment failed? Hit Resume deployment in the portal; it’s idempotent.
- Arc registration wrong tenant? Only fix is to re-image the node. (learn.microsoft.com)
- Logs live under
C:\ClusterStorage\Volume1\Logs
on the resource bridge VM and are downloadable from the portal.
Wrapping Up
By following the four big stages—
- OS install,
- Baseline config,
- Arc registration,
- Portal deployment
You transform commodity servers into a fully Azure-managed edge cloud.
Happy hybrid hacking! 🛠️