Course contentsShow
← Cloud and Infrastructure
Lesson 110 of 2,9892. Compute FoundationsFree lesson

Compute Service Selection Criteria

Choose the right compute service based on workload characteristics, management overhead, and architectural requirements.

Compute Service Selection Criteria

What you'll learn: How to choose the right cloud compute service by evaluating workload needs, management preferences, and architectural goals.

The Decision Framework

You've now learned about VMs, instance types, managed services like AWS Elastic Beanstalk and Azure App Service, and specialized options like spot instances and sole-tenant nodes. But when should you use each one?

Think of compute service selection like choosing transportation: a bicycle, car, bus, or airplane all get you places, but the right choice depends on distance, speed needs, budget, and whether you want to drive yourself or let someone else handle it.

Key Selection Criteria

Workload Characteristics

  • Compute intensity: CPU-heavy tasks favor compute-optimized instances; data processing needs memory-optimized types
  • Traffic patterns: Predictable workloads suit reserved instances; variable loads benefit from auto-scaling services
  • State requirements: Stateless applications work well with managed platforms (App Service, App Engine); stateful workloads may need traditional VMs with persistent storage

Management Overhead

  • Control vs convenience trade-off: Need full OS access? Use VMs. Want zero infrastructure management? Choose managed platforms like Elastic Beanstalk
  • Operational skills: Small teams benefit from higher-level services (PaaS); large DevOps teams can optimize with lower-level control (IaaS)

Architectural Requirements

  • Availability needs: Multi-AZ deployments require VM scale sets or managed services with built-in redundancy
  • Cost constraints: Spot instances for fault-tolerant batch jobs; reserved instances for stable production workloads
  • Compliance: Sole-tenant nodes for regulatory isolation; standard multi-tenant VMs for general use

Key Takeaway: Select compute services by matching workload characteristics (performance, traffic, state) with your tolerance for management complexity and your architectural constraints around availability, cost, and compliance.