---

name: azure-compute description: "Recommend Azure VM sizes, VM Scale Sets (VMSS), and configurations based on workload requirements, performance needs, and budget constraints. No Azure account required — uses public documentation and the Azure Retail Prices API. USE FOR: recommend VM size, which VM should I use, choose Azure VM, VM for web/database/ML/batch/HPC, GPU VM, compare VM sizes, cheapest VM, best VM for workload, VM pricing, cost estimate, burstable/compute/memory/storage optimized VM, confidential computing, VM trade-offs, VM families, VMSS, scale set recommendation, autoscale VMs, load balanced VMs, VMSS vs VM, scale out, horizontal scaling, flexible orchestration. DO NOT USE FOR: deploying VMs or VMSS, deploying apps (use azure-deploy), looking up existing VMs (use azure-resource-lookup), cost optimization of running VMs (use azure-cost-optimization), non-VM services like App Service or AKS."

Azure Compute Skill

Recommend Azure VM sizes, VM Scale Sets (VMSS), and configurations by analyzing workload type, performance requirements, scaling needs, and budget. No Azure subscription required — all data comes from public Microsoft documentation and the unauthenticated Retail Prices API.

When to Use This Skill

• User asks which Azure VM or VMSS to choose for a workload
• User needs VM size recommendations for web, database, ML, batch, HPC, or other workloads
• User wants to compare VM families, sizes, or pricing tiers
• User asks about trade-offs between VM options (cost vs performance)
• User needs a cost estimate for Azure VMs without an Azure account
• User asks whether to use a single VM or a scale set
• User needs autoscaling, high availability, or load-balanced VM recommendations
• User asks about VMSS orchestration modes (Flexible vs Uniform)

Workflow

Use reference files for initial filtering

CRITICAL: then always verify with live documentation from learn.microsoft.com before making final recommendations. If web_fetch fails, use reference files as fallback but warn the user the information may be stale.

Step 1: Gather Requirements

Ask the user for (infer when possible):

Requirement	Examples
Workload type	Web server, relational DB, ML training, batch processing, dev/test
vCPU / RAM needs	"4 cores, 16 GB RAM" or "lightweight" / "heavy"
GPU needed?	Yes → GPU families; No → general/compute/memory
Storage needs	High IOPS, large temp disk, premium SSD
Budget priority	Cost-sensitive, performance-first, balanced
OS	Linux or Windows (affects pricing)
Region	Affects availability and price
Instance count	Single instance, fixed count, or variable/dynamic
Scaling needs	None, manual scaling, autoscale based on metrics or schedule
Availability needs	Best-effort, fault-domain isolation, cross-zone HA
Load balancing	Not needed, Azure Load Balancer (L4), Application Gateway (L7)

Step 2: Determine VM vs VMSS

Workflow:

1. Review VMSS Guide to understand when VMSS vs single VM is appropriate
2. Use the gathered requirements to decide which approach fits best
3. REQUIRED: If recommending VMSS, fetch current documentation to verify capabilities:

   web_fetch https://learn.microsoft.com/en-us/azure/virtual-machine-scale-sets/overview
   web_fetch https://learn.microsoft.com/en-us/azure/virtual-machine-scale-sets/virtual-machine-scale-sets-autoscale-overview
   

4. If web_fetch fails, proceed with reference file guidance but include this warning:

   Unable to verify against latest Azure documentation. Recommendation based on reference material that may not reflect recent updates.

Needs autoscaling?
├─ Yes → VMSS
├─ No
│  ├─ Multiple identical instances needed?
│  │  ├─ Yes → VMSS
│  │  └─ No
│  │     ├─ High availability across fault domains / zones?
│  │     │  ├─ Yes, many instances → VMSS
│  │     │  └─ Yes, 1-2 instances → VM + Availability Zone
│  │     └─ Single instance sufficient? → VM

Signal	Recommendation	Why
Autoscale on CPU, memory, or schedule	VMSS	Built-in autoscale; no custom automation needed
Stateless web/API tier behind a load balancer	VMSS	Homogeneous fleet with automatic distribution
Batch / parallel processing across many nodes	VMSS	Scale out on demand, scale to zero when idle
Mixed VM sizes in one group	VMSS (Flexible)	Flexible orchestration supports mixed SKUs
Single long-lived server (jumpbox, AD DC)	VM	No scaling benefit; simpler management
Unique per-instance config required	VM	Scale sets assume homogeneous configuration
Stateful workload, tightly-coupled cluster	VM (or VMSS case-by-case)	Evaluate carefully; VMSS Flexible can work for some stateful patterns

Warning: If the user is unsure, default to single VM for simplicity. Recommend VMSS only when scaling, HA, or fleet management is clearly needed.

Step 3: Select VM Family

Workflow:

1. Review VM Family Guide to identify 2-3 candidate VM families that match the workload requirements

2. REQUIRED: verify specifications for your chosen candidates by fetching current documentation:

   web_fetch https://learn.microsoft.com/en-us/azure/virtual-machines/sizes/<family-category>/<series-name>
   

   Examples:

   • B-series: https://learn.microsoft.com/en-us/azure/virtual-machines/sizes/general-purpose/b-family
   • D-series: https://learn.microsoft.com/en-us/azure/virtual-machines/sizes/general-purpose/ddsv5-series
   • GPU: https://learn.microsoft.com/en-us/azure/virtual-machines/sizes/gpu-accelerated/nc-family

3. If considering Spot VMs, also fetch:

   web_fetch https://learn.microsoft.com/en-us/azure/virtual-machine-scale-sets/use-spot
   

4. If web_fetch fails, proceed with reference file guidance but include this warning:

   Unable to verify against latest Azure documentation. Recommendation based on reference material that may not reflect recent updates or limitations (e.g., Spot VM compatibility).

This step applies to both single VMs and VMSS since scale sets use the same VM SKUs.

Step 4: Look Up Pricing

Query the Azure Retail Prices API — Retail Prices API Guide

Tip: VMSS has no extra charge — pricing is per-VM instance. Use the same VM pricing from the API and multiply by the expected instance count to estimate VMSS cost. For autoscaling workloads, estimate cost at both the minimum and maximum instance count.

Step 5: Present Recommendations

Provide 2–3 options with trade-offs:

Column	Purpose
Hosting Model	VM or VMSS (with orchestration mode if VMSS)
VM Size	ARM SKU name (e.g., Standard_D4s_v5)
vCPUs / RAM	Core specs
Instance Count	1 for VM; min–max range for VMSS with autoscale
Estimated $/hr	Per-instance pay-as-you-go from API
Why	Fit for the workload
Trade-off	What the user gives up

Tip: Always explain why a family fits and what the user trades off (cost vs cores, burstable vs dedicated, single VM simplicity vs VMSS scalability, etc.).

For VMSS recommendations, also mention:

• Recommended orchestration mode (Flexible for most new workloads)
• Autoscale strategy (metric-based, schedule-based, or both)
• Load balancer type (Azure Load Balancer for L4, Application Gateway for L7/TLS)

Step 6: Offer Next Steps

• Compare reservation / savings plan pricing (query API with priceType eq 'Reservation')
• Suggest Azure Pricing Calculator for full estimates
• For VMSS: suggest reviewing autoscale best practices and VMSS networking

Error Handling

Scenario	Action
API returns empty results	Broaden filters — check armRegionName, serviceName, armSkuName spelling
User unsure of workload type	Ask clarifying questions; default to General Purpose D-series
Region not specified	Use eastus as default; note prices vary by region
Unclear if VM or VMSS needed	Ask about scaling and instance count; default to single VM if unsure
User asks VMSS pricing directly	Use same VM pricing API — VMSS has no extra charge; multiply by instance count

References

• VM Family Guide — Family-to-workload mapping and selection
• Retail Prices API Guide — Query patterns, filters, and examples
• VMSS Guide — When to use VMSS, orchestration modes, and autoscale patterns