AWS Architecture: High Availability, Scaling, and Well-Architected

High availability and fault tolerance

High availability means the system continues operating through failures. Multi-AZ deployments are the foundation.

# High Availability (HA):
# - Minimises downtime during failures
# - Multi-AZ: resources in multiple Availability Zones
# - If one AZ goes down, traffic shifts to healthy AZ

# Fault Tolerance:
# - System CONTINUES operating even during failure
# - Higher bar than HA — zero downtime

# Multi-AZ pattern:
# - ELB (load balancer) spans multiple AZs
# - EC2 instances in each AZ
# - RDS Multi-AZ for automatic DB failover
# - If one AZ fails: traffic routes to healthy AZs automatically

# Availability Zones vs Regions:
# AZ: one or more datacenters within a region
# Region: geographic area with 2+ AZs
# Multiple AZs = HA within a region
# Multiple Regions = DR across regions

Load balancers

Elastic Load Balancing (ELB) distributes traffic across multiple targets.

# Application Load Balancer (ALB):
# - Layer 7 (HTTP/HTTPS)
# - Content-based routing (path, header, query string)
# - Best for microservices, container-based apps
# - Supports WebSockets

# Network Load Balancer (NLB):
# - Layer 4 (TCP/UDP/TLS)
# - Ultra-high performance (millions of requests/sec)
# - Static IP address
# - Best for TCP traffic, gaming, IoT

# Gateway Load Balancer (GWLB):
# - Layer 3 (IP packets)
# - Routes traffic through third-party appliances
# - Best for firewalls, intrusion detection

# Classic Load Balancer (CLB):
# - Legacy, avoid for new designs

# Health checks:
# ELB checks targets regularly
# Unhealthy targets removed from rotation automatically

Auto Scaling

Auto Scaling adjusts the number of EC2 instances based on demand.

# Auto Scaling Group (ASG):
# - Defines min, desired, and max instance count
# - Launches instances from a Launch Template
# - Distributes instances across AZs

# Scaling policies:
# Target tracking: maintain a target metric
#   e.g., keep CPU at 70%

# Step scaling: scale by amount based on metric
#   e.g., add 2 instances if CPU > 80%, add 4 if CPU > 90%

# Scheduled scaling: scale at known times
#   e.g., add 10 instances at 8am on weekdays

# Scale-in protection:
# - Prevent specific instances from being terminated
# - Useful for instances running long jobs

# Cooldown period:
# - Time after scaling before another action
# - Prevents thrashing (rapid scale up/down)

AWS Well-Architected Framework

The six pillars provide best practices for building reliable, secure, cost-efficient AWS applications.

# Six pillars:

# 1. Operational Excellence
#    - Automate operations, learn from failures
#    - Infrastructure as code, CI/CD

# 2. Security
#    - Least privilege, encryption, audit trails
#    - IAM, KMS, CloudTrail, GuardDuty

# 3. Reliability
#    - Recover from failures, scale dynamically
#    - Multi-AZ, backups, Auto Scaling

# 4. Performance Efficiency
#    - Use right resources for workloads
#    - Serverless, caching, right instance types

# 5. Cost Optimisation
#    - Avoid waste, match capacity to need
#    - Reserved Instances, Savings Plans, rightsizing

# 6. Sustainability
#    - Minimise environmental impact
#    - Right-sizing, serverless, managed services

Exam tip

For any architecture question, think: multi-AZ for HA, ALB for HTTP load balancing, Auto Scaling for elasticity, and Multi-AZ RDS for database redundancy. The Well-Architected Framework pillars are always a final exam section.