🔹 Feature: Sidecar Containers in Azure App Service
🔹 What It Does: App Service now supports running your main application container alongside up to 8 additional companion (sidecar) containers in the same instance. All containers share the same network namespace — they communicate over localhost — and start, stop, and scale together.
The pattern:
App Service
├── main container: your application
├── sidecar: Redis (in-instance cache)
├── sidecar: OpenTelemetry Collector (metrics/traces exporter)
└── sidecar: local AI model (e.g. Phi-4)
What Is It Giving You:
✅ Up to 9 Containers Per Instance: 1 main + up to 8 sidecars. Configure via Azure Portal, ARM, Bicep, Terraform, or Azure CLI.
✅ Shared Network Namespace — Localhost Between Containers: Sidecars are addressable from the main container over localhost:
✅ Real Use Cases Without Extra Infrastructure:
• Local caching (Redis, Memcached)
• Observability (OpenTelemetry Collector, Fluentd)
• Local AI inference (Phi-4, small LLMs, embeddings)
• Envoy / NGINX reverse proxy
• Custom auth proxies or credential brokers
✅ Any Container Registry: Docker Hub, Azure Container Registry, or any OCI-compatible registry. Each sidecar has its own image, tag, environment vars, startup command, and port config.
✅ Managed Identity Support: Sidecars can use system- or user-assigned identities to pull private images — no admin credentials stored in config.
✅ Unified Lifecycle: All containers start, stop, restart, and scale together. One App Service Plan, one deployment slot, one monitoring surface.
✅ Available on Linux App Service: Premium v3 / Isolated v2 tiers. Windows sidecar support is in progress.
🌐 https://learn.microsoft.com/en-us/azure/app-service/configure-sidecar
