Azure Functions is Microsoft's serverless compute service on Azure, enabling event-driven execution of code without managing infrastructure. Functions can be triggered by dozens of event sources—HTTP requests, timers, queues, blobs, databases, and more—and support C#, Java, JavaScript/TypeScript, PowerShell, Python, and any HTTP-capable language via Custom Handlers. The platform spans five hosting plans ranging from fully serverless (Flex Consumption, legacy Consumption) to dedicated VMs (App Service), each with different scaling, networking, and pricing trade-offs. Durable Functions extends the model with stateful orchestrations for complex workflows. Understanding hosting plans, the binding model, Durable Functions patterns, language runtimes, deployment tools, and networking is essential for production-grade serverless workloads on Azure.
What This Cheat Sheet Covers
This topic spans 14 focused tables and 130 indexed concepts. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.
Table 1: Hosting Plans
Choose the right hosting plan based on scaling needs, networking requirements, cold-start tolerance, and cost model. Flex Consumption is the recommended serverless option as of 2026.
| Hosting Plan | Example / Key Detail | Description |
|---|---|---|
az functionapp create --plan-type FlexConsumption | • Recommended serverless plan (2024+). Linux only • Per-function scaling, always-ready instances, on-demand billing per execution • Instance sizes: 512 MB/0.25 vCPU, 2048 MB/1 vCPU, 4096 MB/2 vCPU • One app per plan • Supports private endpoints and VNet integration | |
EP1 (1 core/3.5 GB), EP2 (2 core/7 GB), EP3 (4 core/14 GB) | • Pre-warmed + always-ready instances eliminate cold starts • VNet integration and private endpoints supported • Up to 100 instances (Windows) / 20–100 (Linux). Supports deployment slots • Billed per core-second regardless of idle | |
Basic, Standard, Premium App Service SKUs | • Functions run on dedicated VMs with manual or autoscale • Always-On setting prevents host from going idle • No per-execution billing—pay for VM uptime • Best when existing App Service plan is underutilized |