9) Deploying and Managing Infrastructure at Scale

AWS CloudFormation

  • Overview and Core Concepts:

    • Declarative infrastructure modeling: CloudFormation provides a declarative mechanism for specifying, configuring, and provisioning AWS cloud resources using code templates formatted in JSON or YAML.

    • Automated provisioning and orchestration: Resources outlined in a template are provisioned in the exact dependency order required with the precise configurations specified, avoiding manual coordination.

    • Example template workflow:

    • Specification of a Security Group.

    • Provisioning of two Amazon Elastic Compute Cloud (EC2) instances associated with that Security Group.

    • Creation of an Amazon Simple Storage Service (S3) bucket.

    • Deployment of an Elastic Load Balancer (ELB) positioned in front of the two EC2 instances.

  • Infrastructure as Code (IaC) Benefits:

    • Complete administrative control: Eliminates manual resource creation in the AWS Management Console, preventing configuration drift.

    • Versioning and peer review: Changes to cloud infrastructure are codified and reviewed through standard code review and version control processes prior to deployment.

  • Cost Control and Financial Optimization:

    • Granular cost tracking: Every individual resource provisioned within a CloudFormation stack is automatically tagged with a stack identifier, enabling precise stack-level billing visibility.

    • Upfront estimation: Resource costs can be modeled and estimated prior to deployment using template properties.

    • Development savings strategy: Stacks deployed in development environments can be configured for automated deletion at 5 PM5\text{ PM} and safely re-created at 8 AM8\text{ AM}, eliminating idle running costs overnight.

  • Operational Productivity and Reusability:

    • On-demand lifecycle management: Environments can be instantiated, destroyed, and rebuilt rapidly on the fly across multiple AWS Regions and Accounts.

    • Visual architecture generation: Templates automatically generate visual infrastructure diagrams.

    • Declarative simplicity: Eliminates the complexity of manual orchestration, ordering, or custom script writing.

    • Community and vendor resources: Pre-built templates and official AWS documentation can be leveraged directly without re-engineering standard architectural patterns.

  • Broad Resource Coverage and Custom Resources:

    • Extensibility: Supports virtually all AWS resource types natively.

    • Custom Resources: For services, third-party tooling, or edge configurations not directly supported, Custom Resources can be defined (often backed by AWS Lambda) to handle provisioning logic.

  • CloudFormation with AWS Infrastructure Composer:

    • Integration: AWS Infrastructure Composer visually illustrates the resources and connections configured within a CloudFormation stack.

    • Visualization: Diagrams illustrate component relationships, such as security groups, compute launch configurations, database instances, target groups, and load balancers.

WordPress CloudFormation Stack in AWS Infrastructure Composer

AWS Cloud Development Kit (CDK)

  • Core Architecture:

    • Definition using expressive languages: Enables cloud infrastructure to be authored using general-purpose programming languages, including JavaScript/TypeScript, Python, Java, and .NET.

    • Synthesis process: The CDK code compiles and synthesizes down into standard CloudFormation JSON or YAML templates via the CDK Command Line Interface (CLI).

    • Unified delivery: Permits infrastructure code and application runtime code to coexist within the same repository and be deployed simultaneously.

    • Primary use cases: Especially effective for defining serverless architectures (such as AWS Lambda functions) and container orchestrations (such as Docker containers on Amazon ECS or Amazon EKS).

AWS CDK Architecture Workflow
  • TypeScript CDK Implementation Example:

    • Construct stack definition:

    • An extension of core.Stack defines cloud resources programmatically.

    • Virtual Private Cloud (VPC) definition: new ec2.Vpc(this, "MyVpc", { maxAzs: 3 }); defines a VPC spanning up to 33 Availability Zones (where the default is all Availability Zones within the region).

    • Amazon Elastic Container Service (ECS) Cluster: new ecs.Cluster(this, "MyCluster", { vpc: vpc }); provisions an ECS cluster inside the designated VPC.

    • Higher-level load-balanced Fargate service construct:

    • Construct: new ecs_patterns.ApplicationLoadBalancedFargateService(...)

    • Cluster reference: cluster: cluster (Required parameter).

    • CPU allocation: cpu: 512 (Overrides default of 256256 CPU units).

    • Desired task count: desiredCount: 6 (Overrides default of 11 task).

    • Container image specifications: taskImageOptions: { image: ecs.ContainerImage.fromRegistry(...) }

    • Memory limits: memoryLimitMiB: 2048 (Overrides default of 512 MiB512\text{ MiB}).

    • Public exposure: publicLoadBalancer: true (Overrides default of false).

Three-Tier Web Application Architecture

  • Architecture Blueprint:

    • Presentation Tier (Routing): Multi-Availability Zone Elastic Load Balancing (ELB) receives external user traffic and routes incoming requests across compute instances.

    • Logic Tier (Compute): EC2 compute instances (e.g., M5 instance types) running inside an Auto Scaling group distributed across Availability Zone 1, Availability Zone 2, and Availability Zone 3 to maintain fault tolerance and high availability.

    • In-Memory Caching Tier: Amazon ElastiCache clusters positioned alongside the compute layer to store and retrieve active session states and cached operational data.

    • Database Tier (Persistence): Amazon Relational Database Service (Amazon RDS) handling relational read and write transactions with multi-AZ replication.

Typical 3-Tier Web Application Architecture on AWS
  • Common Developer Operational Hurdles:

    • Infrastructure management overhead: Manual configuration of virtual servers, storage volumes, and network security groups.

    • Code deployment friction: Coordinating code artifacts across multiple server targets consistently.

    • Database and networking integration: Manually wiring load balancers, caching layers, and database endpoints.

    • Scaling mechanics: Engineering auto-scaling triggers and policies independently.

    • Standardization: Achieving parity across distinct development, staging, and production environments when the architectural topology (ALB + ASG) remains fundamentally identical.

AWS Elastic Beanstalk

  • Service Overview:

    • Developer-centric abstraction: Acts as a Platform as a Service (PaaS) to simplify application deployments on AWS without requiring low-level infrastructure engineering.

    • Composed AWS primitives: Employs standard foundational components—EC2 instances, Auto Scaling groups (ASG), Elastic Load Balancing (ELB), and Amazon RDS.

    • Consolidated visibility: Bundles all constituent application resources into an integrated dashboard view.

    • Configuration control: Offers full underlying administrative access and configuration override capability if custom modifications are needed.

    • Pricing structure: The Elastic Beanstalk orchestration service is completely free of charge; billing is assessed solely for the underlying provisioned resources (EC2, ELB, RDS, etc.).

  • Managed Operational Responsibilities:

    • Managed by Elastic Beanstalk:

    • Operating system selection, patching, and base instance configuration.

    • Deployment strategies and version rollout execution.

    • Capacity provisioning and scaling adjustments.

    • Load balancing and Auto Scaling policy binding.

    • Health monitoring, responsiveness tracking, and metric reporting.

    • Responsibility of the Developer:

    • Authoring and packaging the application source code.

  • Architecture Deployment Models:

    • Single Instance Deployment: Provisions a single EC2 instance with an Elastic IP; best suited for cost-sensitive development and testing environments.

    • Load Balancer + Auto Scaling Group (LB + ASG): Deploys instances behind a managed load balancer across multiple zones; recommended for production and pre-production web applications requiring resilience.

    • Auto Scaling Group Only (ASG Only): Deploys an ASG without an associated load balancer; optimized for non-web backend workers, asynchronous batch systems, and queue processors.

  • Supported Platform Runtimes:

    • Programming runtimes: Go, Java SE, Java with Tomcat, .NET on Windows Server with IIS, Node.js, PHP, Python, and Ruby.

    • Custom tooling: Packer Builder.

    • Container runtimes: Single Container Docker, Multi-Container Docker, and Preconfigured Docker.

  • Health Monitoring and Diagnostics:

    • Metric collection: A specialized host health agent collects server and application metrics and publishes data directly to Amazon CloudWatch.

    • Enhanced health metrics:

    • Real-time reporting on system state transitions (Ok, Info, Warning, Severe).

    • Network HTTP status distribution (percentages of 2xx, 3xx, 4xx, and 5xx responses).

    • Detailed latency percentiles (P10\text{P10}, P50\text{P50}, P75\text{P75}, P90\text{P90}, and P99\text{P99} latency benchmarks).

    • Host-level performance indicators: CPU utilization breakdown (user, system, idle, I/O wait), and system load averages over 1-minute and 5-minute intervals.

AWS Elastic Beanstalk Enhanced Health Overview

AWS CI/CD and Developer Tools

  • AWS CodeCommit:

    • Managed source control: A fully managed, highly available source-control service hosting private Git repositories.

    • Cloud integration: Direct integration with AWS Identity and Access Management (IAM) for granular access permissions and governance.

    • Collaboration tools: Supports pull requests, branch protection, commit reviews, and automatic version tracking.

  • AWS CodeBuild:

    • Serverless continuous integration: A fully managed build and continuous integration service eliminating the need to provision dedicated build servers.

    • Build workflow: Pulls source code from a repository (such as CodeCommit or GitHub), compiles source code, executes automated testing suites, and packages deployable artifacts.

    • Scalability and pricing: Scales concurrently to meet burst build demands; pay-as-you-go model charges strictly for the compute minutes consumed during the build execution.

  • AWS CodeDeploy:

    • Automated application deployment: Automates application release processes to target hosts.

    • Hybrid architecture: Deploys code seamlessly to Amazon EC2 instances as well as on-premises physical or virtual servers.

    • Deployment agent: Target servers must be provisioned in advance and run the CodeDeploy Agent locally.

    • Version transitions: Manages rolling in-place and blue/green upgrades across fleets (e.g., transitioning fleets from version v1 to v2).

  • AWS CodePipeline:

    • End-to-end orchestration: Orchestrates the sequential stages of Continuous Integration and Continuous Delivery (CI/CD) pipelines.

    • Standard stage lifecycle: Code →\rightarrow Build →\rightarrow Test →\rightarrow Provision →\rightarrow Deploy.

    • Ecosystem integration: Integrates with AWS native services (CodeCommit, CodeBuild, CodeDeploy, Elastic Beanstalk, CloudFormation), third-party repositories (GitHub), and custom plugins.

  • AWS CodeArtifact:

    • Dependency and artifact repository: A fully managed artifact repository service designed for storing, publishing, and retrieving software package dependencies securely.

    • Package manager compatibility: Native support for common packaging ecosystems including Maven, Gradle, npm, yarn, twine, pip, and NuGet.

    • Pipeline integration: Integrates with local development environments and CI/CD tools like AWS CodeBuild to pull authorized dependencies and publish build outputs.

AWS Systems Manager (SSM)

  • Architecture and Scope:

    • Fleet operations at scale: Provides centralized operational control and visibility across Amazon EC2 instances, on-premises virtual machines, and edge devices.

    • Hybrid operational management: Operates as a hybrid tool spanning both cloud and on-premises environments across a suite of 10+10+ distinct management products.

    • OS platform support: Linux, Windows, macOS, and Raspberry Pi OS (Raspbian).

  • The SSM Agent:

    • Prerequisite: Systems must have the SSM Agent installed and active to communicate with the service.

    • Distribution: Pre-installed by default on Amazon Linux AMIs and select Ubuntu AMIs.

    • Operational diagnostic rule: If an instance fails to respond or cannot be managed by Systems Manager, the primary root cause is typically a missing, halted, or misconfigured SSM Agent.

  • Core Capabilities:

    • Run Command: Executes configuration scripts and administrative commands across an entire fleet simultaneously without requiring direct interactive login.

    • Patch Automation: Automates OS and security patch installations to enforce compliance baselines.

  • SSM Session Manager:

    • Browser and CLI shell access: Establishes secure, interactive shell sessions into EC2 instances and on-premises hosts.

    • Perimeter security benefits: Eliminates the requirement for open inbound ports, entirely bypassing the need for port 2222.

    • Bastion and key elimination: Removes the operational overhead of maintaining SSH keys, bastion hosts, and exposed jump boxes.

    • Access governance: Session authorization is strictly controlled via IAM policies.

    • Auditability and compliance: Session logs can be automatically routed and stored in Amazon S3 buckets or CloudWatch Logs for security audits.

  • SSM Parameter Store:

    • Centralized configuration store: Manages key-value pairs representing application configurations, secrets, licensing strings, and database connection strings.

    • Operational characteristics: Serverless, highly durable, scalable, and accessible via standard AWS SDKs.

    • Security tiers: Stores variables as plaintext strings or encrypted SecureString parameters using AWS Key Management Service (AWS KMS).

    • Change tracking: Tracks revisions with built-in version histories and restricts access through IAM permissions.

Infrastructure Deployment and Developer Services Comparison

  • Deployment Services Matrix:

    • AWS CloudFormation:

    • Scope: AWS environments only.

    • Model: Infrastructure as Code (IaC) utilizing declarative templates (JSON/YAML).

    • Coverage: Covers virtually all AWS resources; allows repeatable rollouts across regions and accounts.

    • AWS Elastic Beanstalk:

    • Scope: AWS environments only.

    • Model: Platform as a Service (PaaS).

    • Coverage: Deploys code onto pre-configured web stacks (ALB + EC2 + RDS) for specific runtimes and Docker.

    • AWS CodeDeploy:

    • Scope: Hybrid (AWS EC2 instances and on-premises servers).

    • Model: Automated software deployment agent.

    • Coverage: Deploys and updates application binaries and packages on pre-provisioned operating systems.

    • AWS Systems Manager:

    • Scope: Hybrid (AWS EC2 instances, on-premises servers, and edge devices).

    • Model: Fleet management and operations suite.

    • Coverage: Configures systems, automates patching, provides remote shell access, and runs commands across fleets at scale.

  • Developer Pipeline Services Matrix:

    • AWS CodeCommit: Manages source code inside secure, private, highly available Git repositories.

    • AWS CodeBuild: Serverless build and testing service that compiles code and outputs ready-to-deploy packages.

    • AWS CodeDeploy: Executes automated software deployments and upgrades across servers.

    • AWS CodePipeline: Orchestrates the comprehensive CI/CD pipeline lifecycle from source through deployment.

    • AWS CodeArtifact: Stores and controls access to software package dependencies and internal artifacts.

    • AWS CDK: Provides high-level object-oriented programming abstractions that synthesize down into CloudFormation templates.