Overview
AWS pricing is built on a consumption-based model where you pay only for what you use, with multiple pricing options to optimize cost based on your workload needs. This guide covers the three fundamental cost drivers (compute, storage, outbound data transfer), the tools AWS provides to monitor and control spending, and strategies to reduce your AWS bill. Mastering these concepts is essential for the Cloud Practitioner exam and real-world cloud financial management.
---
Core Pricing Models
The Foundation of AWS Pricing
AWS pricing philosophy is pay-as-you-go — no upfront costs, no long-term contracts required, and spending scales directly with consumption.
The Three Fundamental Cost Drivers:
• Compute – The processing power you use (EC2, Lambda, etc.)
• Storage – The data you store (S3, EBS, RDS, etc.)
• Outbound Data Transfer (Egress) – Data leaving AWS to the internet
> Note: Inbound (ingress) data transfer into AWS is free. You are charged for data going out.
---
EC2 Pricing Models Compared
| Pricing Model | Discount vs. On-Demand | Commitment | Best Use Case |
|---|---|---|---|
| On-Demand | Baseline | None | Unpredictable, short-term workloads |
| Reserved Instances | Up to 75% | 1 or 3 years | Steady-state, predictable workloads |
| Savings Plans | Up to 72% | 1 or 3 years | Flexible commitment across services |
| Spot Instances | Up to 90% | None | Fault-tolerant, interruptible workloads |
| Dedicated Hosts | None (premium) | Optional | Licensing compliance requirements |
Pricing Model Deep Dives
#### On-Demand Instances
• Linux: billed per second (60-second minimum)
• Windows: billed per hour
• No upfront cost or commitment required
#### Reserved Instances (RIs)
• Commit to a 1- or 3-year term for up to 75% savings
• Can be Standard (less flexible, deeper discount) or Convertible (can change instance attributes)
• Unused RI capacity can be shared across accounts in an AWS Organization
#### Savings Plans
Two types to know:
• Compute Savings Plans – Most flexible; applies across any EC2 instance family, region, OS, tenancy, Fargate, and Lambda
• EC2 Instance Savings Plans – Locked to a specific instance family and region; offers deeper discounts than Compute Savings Plans
#### Spot Instances
• Bid on unused AWS capacity
• Can be interrupted by AWS with a 2-minute warning when capacity is reclaimed
• Ideal for: batch processing, data analysis, CI/CD pipelines, rendering, and other fault-tolerant, stateless workloads
#### Dedicated Hosts
• Physical servers fully dedicated to your use
• Most expensive EC2 option — you pay for the entire physical server
• Used primarily for software licensing compliance (e.g., Windows Server, SQL Server per-socket/per-core licensing)
---
AWS Free Tier
The Free Tier allows new and existing customers to use certain services at no charge. There are three distinct types:
| Type | Description | Example |
|---|---|---|
| Always Free | Never expires, available to all customers | Lambda: 1M requests/month |
| 12 Months Free | Available for 12 months from account signup date | EC2 t2.micro: 750 hrs/month |
| Trials | Short-term, service-specific free trials | Amazon Inspector, Lightsail |
Key Terms – Core Pricing Models
• Pay-as-you-go – Consumption-based billing with no upfront requirements
• On-Demand – Hourly/per-second pricing with no commitment
• Reserved Instances (RIs) – Discounted pricing in exchange for 1 or 3-year commitment
• Spot Instances – Discounted, interruptible capacity from AWS's unused pool
• Dedicated Hosts – Physically isolated servers for licensing compliance
• Egress – Outbound data transfer (charged); Ingress = inbound (free)
⚠️ Watch Out For
• Ingress is free; egress is not. The exam often tests whether you know which direction of data transfer incurs charges.
• Spot Instances can be interrupted — never use them for workloads that cannot tolerate interruption (e.g., databases, critical web servers).
• Compute Savings Plans ≠ EC2 Instance Savings Plans. Know that Compute plans cover Lambda and Fargate too; EC2 plans are narrower but offer deeper discounts.
• Dedicated Hosts vs. Dedicated Instances — Dedicated Hosts give you control over the physical server; Dedicated Instances are isolated at the instance level but do not guarantee physical server placement.
---
Cost Management Tools
Tools Overview
| Tool | Primary Purpose | Key Action |
|---|---|---|
| AWS Pricing Calculator | Estimate costs before building | Pre-deployment planning |
| AWS Cost Explorer | Analyze and visualize historical and forecasted costs | Post-deployment analysis |
| AWS Budgets | Set alerts and automated actions on spending limits | Governance and control |
| AWS Trusted Advisor | Broad best-practice checks including cost optimization | Recommendations across 5 pillars |
| AWS Compute Optimizer | ML-based right-sizing recommendations | Resource optimization |
| Cost and Usage Report (CUR) | Most granular billing data available | Deep analysis and reporting |
---
Tool Deep Dives
#### AWS Pricing Calculator
• Estimates cost of planned architecture before you build it
• Helps with budgeting and architectural cost comparisons
• Available at calculator.aws
#### AWS Cost Explorer
• Visualizes historical and forecasted cost and usage data
• Allows filtering by service, account, tag, region, and more
• Provides Right-Sizing Recommendations for EC2 instances
• Forecasting capability based on historical trends
#### AWS Budgets
• Set custom cost, usage, or reservation budgets
• Triggers when thresholds are exceeded or forecasted to be exceeded
• Can send SNS notifications or trigger automatic actions (e.g., restrict IAM permissions, stop EC2 instances)
• Different from Cost Explorer: Budgets is proactive; Cost Explorer is analytical
#### AWS Trusted Advisor
• Provides best-practice checks across 5 categories:
1. Cost Optimization
2. Performance
3. Security
4. Fault Tolerance
5. Service Limits
• Cost optimization checks identify: idle resources, underutilized instances, RI purchase opportunities
• Full check set requires Business or Enterprise Support plan
• Basic and Developer plans: access only to core (limited) checks
#### AWS Compute Optimizer
• Uses machine learning to analyze actual workload metrics
• Recommends optimal configurations for: EC2, Lambda, EBS volumes, ECS on Fargate, Auto Scaling groups
• More precise than Trusted Advisor for resource right-sizing
• Differs from Trusted Advisor: focused specifically on compute resource optimization using historical performance data
#### Cost and Usage Report (CUR)
• The most comprehensive billing dataset in AWS
• Granular line-item data for every AWS service
• Data can go down to the hourly level
• Delivered to Amazon S3 for analysis (e.g., with Athena or QuickSight)
#### Cost Allocation Tags
• Key-value labels attached to AWS resources (e.g., `Project: WebApp`, `Env: Production`)
• Must be activated in the Billing console to appear in cost reports
• Enable cost tracking by: project, department, environment, team, or any custom category
• Essential for chargeback and showback reporting in multi-team environments
Key Terms – Cost Management Tools
• Cost Explorer – Historical and forecasted cost visualization tool
• AWS Budgets – Proactive threshold-based alerting and automated action tool
• Trusted Advisor – Multi-category best-practice recommendation service
• Compute Optimizer – ML-powered compute right-sizing recommendation service
• Cost and Usage Report (CUR) – Most granular billing dataset, delivered to S3
• Cost Allocation Tags – Labels that enable cost categorization in billing reports
• Right-sizing – Selecting the most cost-effective resource size that meets performance needs
⚠️ Watch Out For
• Cost Explorer ≠ AWS Budgets. Explorer analyzes what has happened; Budgets alerts you about what is or might happen.
• Trusted Advisor full checks require Business or Enterprise Support — Basic/Developer plans have very limited access.
• Tags must be activated in the Billing console to appear in cost reports — simply tagging resources is not enough.
• Compute Optimizer ≠ Trusted Advisor — Compute Optimizer uses ML and actual performance data; Trusted Advisor uses simpler threshold-based rules.
---
Billing & Account Management
AWS Organizations & Consolidated Billing
AWS Organizations allows central management of multiple AWS accounts with two key billing benefits:
1. Single Monthly Invoice – All member accounts rolled into one bill
2. Volume Discount Aggregation – Combined usage across accounts helps reach higher discount tiers (e.g., S3 tiered pricing)
3. Reserved Instance & Savings Plan Sharing – Unused discounts from one account automatically apply to matching usage in other accounts
> Structure: One Management (Payer) Account + multiple Member Accounts = an Organization
Billing Console & Key Resources
The AWS Billing and Cost Management Console (Billing Dashboard) provides access to:
• Invoices and billing history
• Payment methods
• Cost reports and Cost Explorer
• Budgets configuration
• Billing preferences and alerts
AWS Cost and Usage Report (CUR)
• Most detailed billing data available
• Line-item level, down to hourly granularity
• Delivered to Amazon S3
• Analyze with Amazon Athena, Amazon QuickSight, or third-party tools
AWS Billing Conductor
• Designed for AWS Partners and large enterprises
• Allows creation of custom (pro-forma) billing data
• Define custom pricing rules and allocate costs across customers or internal business units
• Does not change actual AWS charges — creates a separate billing view
Support Plan Impact on Billing Tools
| Support Plan | Trusted Advisor Access |
|---|---|
| Basic | Core checks only (~7 checks) |
| Developer | Core checks only |
| Business | Full check set (all categories) |
| Enterprise | Full check set + concierge support |
Key Terms – Billing & Account Management
• AWS Organizations – Service for centrally managing multiple AWS accounts
• Consolidated Billing – Feature that combines billing across accounts for single invoice and volume discounts
• Management Account – The payer account in an AWS Organization
• Billing Dashboard – Console where invoices, payment methods, and cost reports are accessed
• CUR (Cost and Usage Report) – Most granular AWS billing dataset
• AWS Billing Conductor – Custom billing tool for partners and enterprises
⚠️ Watch Out For
• Consolidated Billing is a feature of AWS Organizations, not a standalone service.
• RI sharing is automatic by default within an Organization but can be disabled per account.
• Billing Conductor does not alter your actual AWS bill — it only creates custom views/reports.
• The Management Account pays all charges for member accounts in an Organization.
---
Service-Specific Pricing
Amazon S3
Pricing components:
• Storage – Amount of data stored (per GB/month), varies by storage class
• Storage Class – Standard, Intelligent-Tiering, Glacier, etc. (different price points)
• Requests – GET, PUT, COPY, POST, LIST requests are charged separately
• Data Transfer Out – Data leaving S3 to the internet (egress)
Amazon EC2
• On-Demand: Per second (Linux, 60-sec minimum) or per hour (Windows)
• Charges also apply for associated EBS storage, Elastic IPs, and data transfer
Amazon RDS
Pricing components:
• DB Instance hours – Running time of the database instance
• Storage – Provisioned GB/month
• I/O requests – For magnetic (standard) storage type
• Data transfer out – Egress charges apply
• Backup storage – Free up to the size of your DB; charged beyond that
AWS Lambda
Pricing components:
• Requests – Number of function invocations
• Duration – Measured in GB-seconds (memory allocated × execution time in seconds)
• Free Tier: First 1 million requests and 400,000 GB-seconds per month, every month (Always Free)
Data Transfer Pricing Rules
| Transfer Type | Cost |
|---|---|
| Inbound to AWS (ingress) | Free |
| Within the same Availability Zone | Free |
| Between AZs in the same Region | Charged |
| Between AWS Regions | Charged |
| Outbound to the internet | Charged |
Key Terms – Service-Specific Pricing
• GB-seconds – Lambda duration billing unit (memory × time)
• Provisioned storage – RDS and EBS charge for allocated capacity, not just used capacity
• Storage class – S3 tier that determines cost and retrieval characteristics
• I/O requests – Read/write operations charged on some storage types
⚠️ Watch Out For
• Lambda has a permanent Always Free tier — 1M requests and 400K GB-seconds never expire.
• RDS charges for provisioned storage, not just data used — even an empty DB incurs storage costs.
• Data transfer between AZs costs money — architecting within a single AZ avoids this but reduces availability.
• S3 charges per request type — high-frequency GET requests can significantly increase bills.
---
Savings Strategies
Right-Sizing
Right-sizing = Matching resource size to actual workload needs — not too large (over-provisioned) and not too small (under-provisioned).
• Use AWS Compute Optimizer for ML-based recommendations
• Use Cost Explorer's right-sizing feature for EC2 suggestions
• Review Trusted Advisor cost optimization checks regularly
• Downsize or terminate idle and underutilized resources
Choosing the Right Pricing Model
Decision Framework:
```
Is the workload steady and predictable?
→ YES: Reserved Instances or Savings Plans
Is the workload flexible and fault-tolerant?
→ YES: Spot Instances
Is the workload short-term or unpredictable?
→ YES: On-Demand
Do you have software licensing requirements?
→ YES: Dedicated Hosts
```
Reserved Instance Sharing in Organizations
• Enabled by default in Consolidated Billing
• Unused RI capacity automatically applies to matching usage in any member account
• Creates significant savings for organizations where some accounts have surplus reservations
• Can be disabled per account if isolation is required
Spot Instance Best Practices
Good fit for Spot Instances:
• Batch processing jobs
• Data analysis and big data
• CI/CD and testing pipelines
• Video rendering and encoding
• Machine learning training (with checkpointing)
• Containerized microservices with Auto Scaling
Bad fit for Spot Instances:
• Production databases
• Critical web servers with strict uptime requirements
• Any stateful, non-interruptible workloads
Key Terms – Savings Strategies
• Right-sizing – Optimizing resource configurations to match actual workload requirements
• Over-provisioning – Allocating more resources than needed, leading to wasted spend
• RI sharing – Feature allowing unused Reserved Instance discounts to apply across Organization accounts
• Fault-tolerant – Workloads designed to handle unexpected interruptions without data loss or failure
⚠️ Watch Out For
• Savings Plans apply automatically to matching usage — you don't assign them to specific instances.
• Right-sizing is an ongoing process, not a one-time task — workloads change over time.
• Spot Instances are not suitable for all cost-cutting scenarios — always evaluate interruption tolerance first.
• Committing too aggressively to RIs can be worse than On-Demand if your workload shrinks — consider Convertible RIs for more flexibility.
---