I'm starting a greenfield project with about 15 microservices. I want to minimize management overhead but need high scalability. AKS seems powerful but complex, while Container Apps (ACA) looks simpler. Which one handles auto-scaling and Dapr integration better for a growing startup?
3 answers
For a startup, Azure Container Apps (ACA) is usually the smarter choice because it is serverless and scales to zero, saving you money during idle times. ACA is built on K8s but abstracts the complexity, and it has native Dapr support baked in, making service-to-service communication and state management incredibly simple. However, if you need custom admission controllers or specific GPU configurations for AI workloads, AKS is the way to go. AKS gives you full cluster control, but be prepared for the "Day 2" operations overhead like patching and manual node pool scaling.
Are you planning to use a service mesh like Istio, or do you prefer the built-in ingress and traffic splitting features offered by ACA?
If you don't have a dedicated DevOps person, go with Azure Container Apps. AKS will swallow your time with cluster management and updates.
Totally agree with Emily. ACA is basically "Kubernetes for humans" and fits the startup "move fast" mentality perfectly.
We actually want to avoid managing a service mesh if possible. We’ve heard that ACA handles blue-green deployments and traffic splitting natively through its revision management system. Does this mean we can skip learning complex Helm charts for deployments? We really want to focus on our code rather than the YAML boilerplate that usually comes with a full Kubernetes setup.