Jon Rumsey

An online markdown blog and knowledge repository.

Project maintained by nojronatron Hosted on GitHub Pages — Theme by mattgraham

Serverless Intelligent Apps on Azure Container Apps

Series of MSFT Reactor presentations on Serverless, Intelligent Apps, and Azure Container Apps.

Introduction
Azure Container Apps (ACA)
Azure Developer CLI - azd
Deploying ACA with Dapr using azd
Resources
Footer

Introduction

Presenter: Jonah Andersson, Sr. Azure Consultant at Solidify.

Founder/Leader of Azure User Group, Sweden
Fullstack .NET focused developer.
Lead Cloud DevOps Engineer Architect.
Author of Learning Microsoft Azure (O'Reilly).
And much more.

Azure Functions? Not anymore! Now use Azure Serverless:

Solve Queuing (workflow) problems.
Events need to be able to wait on each other.
Large, complex apps can be challenging to keep a manageble, performant workflow.

Serverless + AI + Cloud Native Intelligent Apps.

Serverless Computing

Not really serverless, right?

Is an application development and execution model.
Empowers developers to build and run app code without provisioning or managing servers or backend infrastructure.
There are still servers, but they and their configuration are abstracted for simplification for dev, build, debug, and run.
Other cloud resources are serverless: DBs, Storage, Containers, Microservices, and Cloud Native.

Serverless on Azure

Azure Functions: Event-driven, serverless computer.
Azure Durable Functions: Extension of Azure Functions for stateful, event-driven workflows.
Azure Serverless Databases: Azure SQL Serveless, Azure Cosmos DB Serverless, etc.
Azure Logic Apps: UI User WOrkflows for apps, data, and devices on cloud and on-prem using pre-built Connectors.
Azure Kubernetes (AKS), Azure Container Apps (ACA), etc.

Cloud Native

Cloud Native Computing Foundation defines Cloud Native technologies as:

Containers.
Microservices.
Declarative, cloud-based APIs.
Enable loosely coupled cloud systems.
Robust automation.
High-impact, frequent changes with 'minimal toil'.

MSFT Pillars of Cloud Native:

Cloud Infrastructure: Microsoft Azure PaaS.
Modern Design: 12-factor application, well architected framework for Azure.
Microsofvices: Follows 'process' principle of 12-factor application.
Containers: Code, dependencies, and runtime are packages into a binary "container image" that has Disposability and Concurrency features.
Backing Services: ANy service the app consumes over the network as part of its normal operation.

The 12 Factors

Search for the Twelve-Factor App to learn about:

Codebase
Dependencies
Config
Backing services
Building, release, run
Processes
Port binding
Concurrency
Disposability
Dev/prod parity
Logs
Admin Processes

Monolithic vs Cloud Native

Cloud-native separates Client Apps from Cloud-hosted Docker/Web Apps with backing Micro-Services via an API Gateway to provide features and capability.

Workflow is managed by Event Bus (publish/subscribe channel) behind the microservices so event-based messaging is enabled.

Client App Examples:

Mobile App.
Traditional Web App.
SPA Web App.

Ideal Use Cases

Migration and Modernization of legacy application to the cloud.

Two causes: Forced to by problem or policy, or by implementing cloud-native feature solutions.
Deployment Models: On-prem, IaaS, Containers and PaaS, etc.

Using IaaS enabled "lift and shift" migration method which could be quick, and enables full migration to cloud native as a whole.

Azure Container Apps (ACA)

Serverless, container-based hosting service.

Focuse on apps instead of infrastructure.
Event-driven scale from zero to many.
No-pay when not in use (zero scale).
Fast, dynamic scaling to handle incoming requests.

Build Ideas:

Microservices.
Public API Endpoints.
Web Apps.
Event-driven Processing.
Background Processing.

Auto-Scaling:

Individual microsoervcices can scale indipendently.
Scaling determined by number of concurrnt http requests.
Number of messages in the Queue informs auto scaling feature.
Level of CPU and memory load also drive scaling automation.

Benefits:

Serverless, pay-per-use.
Automatic scaling.
Run containers from any registry (Docker etc) or Azure Container Registry (ACR).
Unified Management: Azure CLI, Portal, etc.
Microservices using Dapr.
Itegration with servless like Azure Functions.
Traffic splitting for blue-green deployments.
Internal ingress and DNS-based services discovery.

ACA can be a Container App Environment:

Container Apps work in concert to act as the primary service.
Other serverless features like Costmos DB, Cache for Redis, Managed Identities, and Azure Service Bus.
Azure App Insights, zure Monitor, and Log Analytics can be plugged-in to the Container App Environment.

Managing Complexity of Microservices:

Dapr does this!
Provides Building blocks and complonents to build microservices.
Hosting can be done by Azure, AWS, Google Cloud, etc.
Dapr: Provides APIs to simplify microsorvice connectivity.
Dapr suports Pub/Sub messaging and Service Invocation models.
Service discovery, message broker integration, encryption, observabilitym, Secrets, and Configuration are managed by Dapr Sidecar.
Dapr Sidecar can be thought of as a daemon that sits between the Container App and the custo Component.
Dapr Sidecar uses Service Bus to manage these communications.

Azure Developer CLI - azd

Is an open-source, developer-centric tool.

Azure CLI is the Administrative version of this.
AZD is focused on Developer needs to build and deploy WebApps.

Building An App For Cloud:

Planning.
Proof of Concept.
Team.
Responsibilities.
Tech Stack selection.
Tools for the team: Dev, Test, CI, Deployment, Debug, etc.

Other considerations:

Frontend platform and type.
Backend platform and type.
DevSecOps and CI-CD
IaC: Infrastructure as Code selection.
Cloud integration and service considerations.
Testing and Automation.

AZD Features:

Developer-centric cloud developent tool for Azure.
Templates to speed development and simplify workflow.
Open-source, ployglot.
Good for Proof-of-Concept projects.

Workflow:

Select azd template.
azd init --template [template name]
Code and configure infra, params, azure.yml.
azd up
Iterate code changes.
azd deploy

Cycle through 5 and 6 through the lifecycle of App development.

azd provision: Optionally update Azure resources by modifying the template Infrastructure as Code (IaC).

azd monitor: Opens Azure AppInsights Metrics locally.

Locally develop:

PowerShell, VSCode plugin.
GitHub Codespaces.
DevContainers.

Deploying ACA with Dapr using azd

This was a live demo and the following are random thoughts while watching the demo:

Using a 'typical azd template' project.
Is a local project that Jonah will push to cloud using azd.
Uses pub-sub via Azure Service Bus.
Can be configured to use DevContainers, or build in GitHub, using dockerfiles and yaml.
azd Templates enabled quick spin-up of Terraform or OpenAI services behind your ACA project.
dapr has a Terminal command-line that enables running the Project with named resources and tcp port.
In the Q&A it was noted that ACA is probably simpler to setup and get running that AKS (Azure Kubernetes Service).
azd up: Does package checkout, build, and creates Containers Endpoint, and logs these activities in the Terminal.