Full-Stack Architect

Brisbane, Australia

January 8, 2026

Technical Guide12 min read

AWS Cloud Costs Explained

A practical guide to understanding where your AWS bill comes from, what's actually costing you money, and how to optimise your cloud spend.

The Pay-As-You-Go Reality

AWS (Amazon Web Services) operates on a pay-as-you-go model. Unlike traditional hosting where you pay a fixed monthly fee regardless of usage, AWS charges based on actual resource consumption. This provides flexibility but also means costs can be unpredictable if you don't understand the billing model.

Based on real-world production deployments, we've identified where costs typically accumulate and the hidden charges that catch most organisations off guard. This guide will help you understand your AWS bill and make informed architecture decisions.

Your Situation Will Be Different

The costs and percentages in this guide are illustrative examples based on typical production workloads. Your actual costs will vary significantly based on your architecture, traffic patterns, data volumes, region, and AWS pricing changes. Use this as a starting point for understanding cost structures — not as a quote or estimate for your specific use case. Always verify current pricing on theAWS Pricing page.

Regional Pricing Matters

For Australian businesses, the Sydney region (ap-southeast-2) is typically 10-20% more expensive than US regions. The trade-off is latency — Sydney provides sub-50ms response times for local users, while US regions add 150-200ms.

Where Your Money Actually Goes

Typical cost distribution for a production web application on AWS

EC2 (Compute)30-35%

Virtual servers running your applications

EKS (Kubernetes)15-20%

Container orchestration platform

Networking10-15%

NAT Gateways, VPC, data transfer

Load Balancers5-8%

ALB/NLB for traffic distribution

RDS (Database)5-10%

Managed relational databases

CloudWatch4-6%

Logging and monitoring

Storage3-5%

S3, EBS, EFS storage

Tax (GST)10%

Australian GST on all services

EC2 Instance Pricing Guide

Sydney region (ap-southeast-2) On-Demand pricing

Instance Type	vCPU	Memory	Hourly	Monthly*	Best For
t2.micro	1	1 GB	$0.015	~$11	Dev/testing, small apps
t3.medium	2	4 GB	$0.053	~$39	Light production workloads
t3.large	2	8 GB	$0.106	~$77	Standard web applications
t3.xlarge	4	16 GB	$0.211	~$154	Medium workloads
m6a.large	2	8 GB	$0.108	~$79	General purpose, balanced
m5.xlarge	4	16 GB	$0.240	~$175	Production applications

$446

Outdated cluster/month

That's $4,464/year extra per cluster just for being behind on updates. Schedule quarterly cluster upgrades to avoid this.

Cost Optimisation Strategies

Practical ways to reduce your AWS bill without sacrificing performance

Strategy	Savings	Effort	Risk	Best For
Spot Instances	60-90%	Medium	Can be interrupted	CI/CD, batch jobs, dev environments, stateless workers
Reserved Instances (1yr)	30-40%	Low	Commitment required	Steady-state production workloads
Reserved Instances (3yr)	60-72%	Low	Long commitment	Core infrastructure, databases
Right-sizing	20-40%	Medium	None	Over-provisioned instances (<30% CPU)
VPC Endpoints	Variable	Low	None	Heavy S3/DynamoDB traffic (avoid NAT costs)
Log Retention Policies	30-50%	Low	None	Reducing CloudWatch log costs

Spot Instances: The 90% Discount Secret

Spot instances use AWS's spare capacity at massive discounts (60-90% off On-Demand). The catch? AWS can reclaim them with 2 minutes notice. But for the right workloads, they're transformative.

✓ Great for:

• CI/CD pipelines and build servers
• Development and testing environments
• Batch processing and data analysis
• Kubernetes worker nodes (with proper PDBs)

✗ Not recommended for:

• Databases or stateful applications
• Single points of failure
• Long-running, uninterruptible jobs
• Applications without auto-recovery

Architecture Patterns & Cost Implications

Typical monthly costs for different deployment architectures

Simple Web Application

$100 - $300/month

Small business website, internal tools, simple APIs

EC2 or LightsailRDS (single AZ)S3CloudFront

Scalable Container App

$1,500 - $5,000/month

SaaS applications, microservices, high-traffic web apps

EKS clusterALBRDS Multi-AZElastiCacheS3

Enterprise Multi-Environment

$5,000 - $20,000+/month

Large organisations with dev/staging/production

Multiple EKS clustersMulti-AZ RDSVPC peeringWAFMultiple ALBs

Monitoring & Controlling Spend

AWS Cost Explorer

Visualise spending patterns, identify trends, and forecast future costs. Enable daily granularity for detailed analysis.

AWS Budgets

Set up alerts at 50%, 80%, and 100% of your budget to catch cost issues before they become problems.

Tagging Strategy

Implement consistent resource tags (Environment, Project, CostCenter) to allocate costs and identify optimisation opportunities.

Key Takeaways

EC2 and container services (EKS) typically account for 40-60% of costs

NAT Gateways and load balancers are fixed costs that add up quickly

CloudWatch log ingestion ($0.67/GB) often exceeds storage costs

EKS extended support can cost 5x the base cluster fee

Spot instances offer 60-90% savings for interruptible workloads

Reserved Instances provide 30-72% savings for steady-state workloads

Data transfer costs are often underestimated — use VPC endpoints

Regular architecture reviews prevent silent cost creep

AWS Cloud Costs Explained

The Pay-As-You-Go Reality

Where Your Money Actually Goes

EC2 Instance Pricing Guide

The Hidden Costs That Catch Everyone

NAT Gateway

EKS Extended Support

CloudWatch Logs

Data Transfer

EBS Snapshots

Idle Load Balancers

The EKS Extended Support Trap