Compute Selection Cheatsheet¶

Use this page to compare Azure compute services quickly during early architecture workshops.

Service	Use Case	Scaling Model	Cost Model	Team Skill Required	Control Level
Virtual Machines	Legacy apps, custom OS dependencies, lift-and-shift	Manual or VMSS	Pay for provisioned instances	Medium to high	Highest
App Service	Web apps, APIs, standard background jobs	Plan scale-up and scale-out	Plan-based	Low to medium	Moderate
Azure Functions	Event-driven, bursty, timer, queue processing	Event-driven elastic scale	Consumption, premium, or dedicated	Low to medium	Low to moderate
Azure Container Apps	Containerized APIs, jobs, microservices without AKS	Revision and event-based scaling	Consumption and dedicated environment options	Medium	Medium
AKS	Complex microservices, platform teams, custom networking	Cluster and pod autoscaling	Cluster plus operations overhead	High	High

Decision guidance¶

Choose VMs when OS-level control or legacy dependencies are non-negotiable. [Documented]
Choose App Service for mainstream web workloads with minimal platform toil. [Documented]
Choose Functions when triggers define execution and idle cost matters. [Observed]
Choose Container Apps when the app is containerized but AKS would be excess operational weight. [Correlated]
Choose AKS only when workload complexity benefits from Kubernetes control more than it suffers from platform overhead. [Inferred]

flowchart TD
    A[Need compute] --> B{Need OS or kernel control?}
    B -->|Yes| C[Virtual Machines]
    B -->|No| D{Container packaging required?}
    D -->|No| E{Event driven?}
    E -->|Yes| F[Azure Functions]
    E -->|No| G[App Service]
    D -->|Yes| H{Need Kubernetes control plane features?}
    H -->|Yes| I[AKS]
    H -->|No| J[Azure Container Apps]

Microsoft Learn references¶

https://learn.microsoft.com/en-us/azure/architecture/guide/technology-choices/compute-decision-tree
https://learn.microsoft.com/en-us/azure/architecture/guide/technology-choices/