Compute Models

Overview

Modern cloud-native systems offer three primary compute models, each optimizing different dimensions of cost, isolation, scalability, and operational complexity. This section explores the architecture, trade-offs, and decision criteria for each.

Models Covered

1. Virtual Machines (VMs) – Full OS isolation, high compatibility, longer startup; suited for legacy applications, strict isolation requirements, and mixed workloads.

2. Containers – Process-level isolation, fast startup, dense packing; the foundation of microservices and Kubernetes-native architectures.

3. Serverless/FaaS – Event-driven, auto-scaling, pay-per-execution; ideal for transactional workloads, event processors, and bursty compute.

---

📄️ Virtual Machines (VMs)

Full OS isolation, live migration, mixed workloads, and legacy application compatibility

📄️ Serverless & Functions-as-a-Service (FaaS)

Event-driven compute, auto-scaling, and pay-per-execution with no infrastructure management

📄️ Containers

Package applications with dependencies using container technology.