AWS Infrastructure for a Highly Available Web Application (Terraform, ALB, Auto Scaling)

I designed and implemented a highly available web application infrastructure on AWS using Terraform (Infrastructure as Code). The solution addressed critical reliability and cost inefficiencies in an existing environment by introducing multi-AZ architecture, load balancing, automated scaling, and scheduled capacity management.

The result was a resilient, self-healing system capable of handling variable traffic patterns while significantly reducing idle compute costs.

Business Context

A client-facing web application was experiencing recurring operational challenges:

• Intermittent downtime during traffic spikes
• No automated scaling strategy
• High infrastructure costs due to 24/7 EC2 provisioning
• Manual and error-prone deployment processes

The environment lacked elasticity and observability, resulting in poor reliability and inefficient resource utilization.

The objective was to redesign the infrastructure into a scalable, highly available, and cost-optimized cloud architecture using Infrastructure as Code.

Technical Objectives

• Architect a fault-tolerant system across multiple Availability Zones
• Implement Infrastructure as Code using Terraform
• Introduce load balancing to distribute traffic efficiently
• Enable automatic scaling based on demand
• Optimize cost through scheduled capacity adjustments
• Eliminate manual infrastructure provisioning and configuration drift

Architecture Overview

High-Level System Design

Diagram

1. Infrastructure as Code (Terraform)

The entire infrastructure was codified using Terraform to ensure:

• Declarative and version-controlled infrastructure
• Repeatable deployments across environments
• Reduced configuration drift
• Faster recovery and rollback capabilities

Core resources included:

• VPC networking components
• Security groups
• Launch templates
• Application Load Balancer
• Target groups
• Auto Scaling Groups
• Scheduled scaling policies

2. High Availability Architecture

The system was deployed across multiple Availability Zones to eliminate single points of failure.

An Application Load Balancer (ALB) was introduced to:

• Distribute traffic across healthy EC2 instances
• Perform continuous health checks
• Improve fault tolerance and service continuity

This significantly improved application resilience under load.

3. Standardized Compute Layer (Launch Templates)

A Launch Template was implemented to enforce consistency across EC2 instances:

• Amazon Linux 2 base image
• Predefined instance configuration (t2.micro)
• Automated bootstrapping via user data
• Security group attachment for controlled HTTP access

This eliminated manual server configuration and ensured consistency at scale.

4. Elastic Scaling Strategy (Auto Scaling Group)

An Auto Scaling Group was configured to manage compute capacity dynamically:

• Minimum capacity: 0
• Desired capacity: 1
• Maximum capacity: 1

This allowed the environment to scale in response to demand while maintaining strict cost controls.

The ASG also ensured self-healing by replacing unhealthy instances automatically.

5. Cost Optimization via Scheduled Scaling

To align infrastructure usage with predictable business hours, scheduled scaling was implemented:

EventTime (UTC)ActionBusiness Start08:00Scale to 1 instanceBusiness End20:00Scale down to 0 instances

This introduced a predictable cost optimization model, eliminating unnecessary idle compute spend.

Key Outcomes

Reliability Improvements

• Eliminated single-instance dependency
• Improved uptime through multi-AZ design and health-checked load balancing
• Introduced automatic recovery via Auto Scaling

Operational Efficiency

• Fully automated provisioning using Terraform
• Removed manual deployment processes
• Reduced configuration inconsistencies

Cost Optimization

• Eliminated 24/7 compute waste
• Introduced time-based scaling aligned to business usage patterns
• Achieved significant reduction in idle resource consumption

Engineering Challenges & Resolution

A key challenge encountered involved IAM permission constraints during EC2 provisioning via Launch Templates. This required:

• Debugging Terraform execution roles and instance profiles
• Reviewing AWS permission boundaries
• Refining least-privilege IAM policies for compute resources

This exercise reinforced the importance of secure-by-design infrastructure provisioning in production environments.

Key Technical Takeaways

• Infrastructure as Code is essential for scalable and repeatable cloud systems
• Load balancing is foundational to designing resilient distributed applications
• Auto Scaling must be aligned with real-world traffic behavior, not static assumptions
• Cost efficiency is achieved through architecture design, not post-deployment optimization
• Production-grade systems require automation, not manual intervention