This chapter covers AWS pricing models, a critical topic for the CLF-C02 exam under Domain 4: Billing, Pricing, and Support (Objective 4.1). Understanding pricing models is essential because cost optimization is a shared responsibility between AWS and the customer, and the exam tests your ability to choose the right model for different workloads. This objective carries approximately 12-16% of the exam weight. By the end of this chapter, you will differentiate between On-Demand, Reserved, Savings Plans, Spot Instances, and Dedicated Hosts, and know when to apply each.
Jump to a section
Imagine you run a small catering business. Your restaurant has a kitchen, tables, and a storage room. In a traditional on-premises setup, you'd buy all the kitchen equipment upfront—ovens, fridges, and utensils—regardless of how many customers you serve. That's like buying servers and data center space before you know your demand. AWS pricing models are like a restaurant with a flexible payment system. First, you pay a monthly membership fee (Reserved Instance) to reserve a table and get a discount on meals. Second, you pay per plate at the buffet (On-Demand) for each customer who walks in, with no upfront commitment. Third, you can bid for a table during off-peak hours (Spot Instance) and get huge discounts, but you might be asked to leave if a full-price customer arrives. Finally, you pay only for the electricity and water you use (Serverless/Pay-per-request), like a kitchen that charges per burner usage per minute. The mechanism: AWS has massive economies of scale—they buy servers in bulk. They pass savings to you when you commit (Reserved) or when you use spare capacity (Spot). On-Demand is the most flexible but most expensive because you pay for the convenience of no commitment. Just like a restaurant, AWS needs to cover fixed costs; your pricing choice determines how much risk you share with AWS.
What Are AWS Pricing Models and Why Do They Exist?
AWS offers multiple pricing models to match different usage patterns and risk profiles. The fundamental principle is that AWS passes on its cost savings from large-scale infrastructure to customers in exchange for commitment or flexibility. The problem they solve: traditional IT procurement requires you to estimate capacity months in advance, leading to either over-provisioning (waste) or under-provisioning (poor performance). AWS pricing models let you align costs with actual usage.
On-Demand Pricing
On-Demand is the simplest model: you pay for compute or database capacity by the hour or second (depending on the service) with no long-term commitments. This is ideal for workloads with unpredictable traffic, such as a new startup testing a product or a batch job that runs sporadically. For EC2, On-Demand pricing varies by instance type, region, and operating system. For example, a t3.micro in US-East-1 costs about $0.0104 per hour. The key exam point: On-Demand is the most expensive per unit but offers maximum flexibility. AWS does not require any upfront payment; you are billed at the end of the month based on actual usage.
Reserved Instances (RIs)
Reserved Instances provide a significant discount (up to 72% compared to On-Demand) in exchange for a commitment to use a specific instance type in a specific region for a 1- or 3-year term. There are three payment options: All Upfront (largest discount), Partial Upfront, and No Upfront (smallest discount). RIs also come in three scope options: Standard (region-specific) and Convertible (allows changing instance attributes). Standard RIs offer a fixed discount and are best for steady-state workloads like a production web server that runs 24/7. Convertible RIs offer less discount but let you change instance family, OS, or tenancy. Exam trap: RIs are not a physical instance; they are a billing discount applied to matching running instances. You still launch instances as usual; the RI discount is automatically applied.
Savings Plans
Savings Plans (SPs) are a flexible pricing model introduced in 2019. Instead of committing to a specific instance type, you commit to a dollar amount of compute spend per hour (e.g., $10/hour) for a 1- or 3-year term. AWS applies the discount to any EC2 instance, Lambda, or Fargate usage up to your commitment. Any usage above the commitment is charged at On-Demand rates. Savings Plans offer similar discounts to RIs but with more flexibility—you can change instance families, regions (within a specific scope), or even use serverless compute. There are two types: Compute Savings Plans (most flexible, apply to any compute across regions) and EC2 Instance Savings Plans (less flexible, apply to a specific instance family in a region). Exam tip: Savings Plans are gradually replacing RIs for many use cases because they simplify management.
Spot Instances
Spot Instances let you bid on unused EC2 capacity at discounts of up to 90% off On-Demand. However, AWS can reclaim the instance with a 2-minute warning if the Spot price exceeds your bid or if capacity is needed for On-Demand customers. Spot Instances are ideal for fault-tolerant, stateless workloads such as big data analytics, containerized batch jobs, or CI/CD pipelines. You can use Spot Instances with Auto Scaling groups and Spot Fleet to maintain capacity. Exam trap: Spot Instances are not suitable for databases or any workload that cannot handle interruptions. The Spot price fluctuates based on supply and demand, but AWS now uses a market-based pricing model where you pay the Spot price (not your bid) when the instance runs. The maximum price you are willing to pay is your bid; if the Spot price exceeds your bid, the instance is terminated.
Dedicated Hosts and Dedicated Instances
Dedicated Hosts provide a physical server fully dedicated to your use, giving you visibility and control over instance placement. This is useful for licensing requirements (e.g., Windows Server per-core licenses) or regulatory compliance. Dedicated Instances run on hardware dedicated to a single customer but may share hardware with other instances from the same account. Both are more expensive than multi-tenant instances. Exam focus: Know the difference—Dedicated Hosts give you server-bound software license portability; Dedicated Instances do not.
Per-Second Billing
For many EC2 instances launched after 2017, AWS bills in one-second increments (minimum 60 seconds) for Linux instances. This reduces costs for short-lived workloads. Windows instances are billed per hour. Per-second billing applies to On-Demand, Reserved, and Spot Instances.
Free Tier
AWS Free Tier includes Always Free offers (e.g., 1 million Lambda requests per month), 12-month free offers (e.g., 750 hours of t2.micro or t3.micro per month), and short-term trials (e.g., Amazon SageMaker 2 months free). Exam questions often ask about Free Tier limits, so memorize the 750 hours per month for EC2 and 5 GB of S3 storage.
Comparison to On-Premises
On-premises: You buy hardware upfront (CapEx), pay for power, cooling, and staff, and bear the risk of underutilization. AWS: You convert CapEx to OpEx, paying only for what you use. Pricing models let you further optimize: On-Demand = no commitment, highest cost; Reserved/SP = commitment, lower cost; Spot = highest risk, lowest cost. The trade-off is always between flexibility and discount.
When to Use Each Model
On-Demand: Unpredictable workloads, short-term projects, development/testing.
Reserved Instances: Steady-state production workloads, known capacity needs for 1-3 years.
Savings Plans: Mix of compute services (EC2, Lambda, Fargate) with predictable baseline spend.
Spot Instances: Fault-tolerant, flexible workloads like batch processing, data analysis, or stateless web servers.
Dedicated Hosts: Licensing or compliance requirements requiring physical isolation.
Choose a pricing model for EC2
Start by analyzing your workload: Is it steady-state or intermittent? Can it tolerate interruptions? For a production web server that runs 24/7 for 3 years, use Reserved Instances or a Savings Plan. For a batch job that runs nightly and can restart, use Spot Instances. For a proof-of-concept with unknown usage, use On-Demand. AWS provides the AWS Pricing Calculator to estimate costs. The exam expects you to match the model to the scenario.
Purchase a Reserved Instance via Console
In the AWS Management Console, navigate to EC2 > Reserved Instances > Purchase Reserved Instances. Select instance type (e.g., t3.medium), scope (Regional or Zonal), term (1 or 3 years), payment option (All Upfront, Partial Upfront, No Upfront), and offering class (Standard or Convertible). After purchase, the RI discount is automatically applied to any matching instance in your account. You do not need to launch a separate instance. Behind the scenes, AWS tracks your RI inventory and applies the discount to the first matching instance in the billing hour. Exam tip: RIs are applied at the account level; if you have multiple accounts consolidated under AWS Organizations, you can share RIs across accounts.
Set up a Savings Plan
Go to AWS Cost Explorer > Savings Plans > Purchase. Choose Compute Savings Plans or EC2 Instance Savings Plans. Enter your hourly commitment (e.g., $5/hour). Select term (1 or 3 years) and payment option. After purchase, the discount applies automatically to eligible compute usage. The Savings Plan covers the first $5/hour of compute; any usage beyond that is billed at On-Demand rates. AWS recommends analyzing your past usage in Cost Explorer to determine the optimal commitment. Exam focus: Savings Plans are more flexible than RIs but require careful commitment sizing to avoid waste.
Request Spot Instances via Auto Scaling
Create an Auto Scaling group and specify a launch template. In the request, set 'Purchase option' to 'Spot'. Optionally, set a maximum price (bid) or leave it as the On-Demand price (recommended). AWS will launch Spot Instances if capacity is available. You can also use a Spot Fleet to request a mix of instance types. Behind the scenes, AWS monitors Spot capacity. If the Spot price exceeds your bid or capacity is reclaimed, the instance receives a 2-minute termination notice. Use lifecycle hooks to gracefully shut down. Exam tip: Spot Instances are best used with instance fleets and multiple Availability Zones to reduce interruption risk.
Evaluate costs with AWS Pricing Calculator
Go to calculator.aws. Add services (e.g., EC2, S3, RDS). Configure instance types, storage, data transfer, and pricing models (On-Demand, Reserved, Spot). The calculator estimates monthly costs and compares models. For example, a t3.medium On-Demand for 730 hours/month costs ~$30; a 1-year All Upfront RI costs ~$190 (saving ~47%). Use this tool to justify pricing model decisions. Exam questions may present cost scenarios; practice calculating savings percentages.
Scenario 1: E-commerce Startup with Variable Traffic
A new e-commerce startup expects low traffic initially but may see spikes during promotions. They use On-Demand EC2 instances for the first 6 months to avoid upfront commitment. As traffic stabilizes, they analyze usage and find that baseline traffic requires 10 t3.medium instances 24/7. They purchase 1-year Standard RIs with Partial Upfront for those 10 instances, saving 40%. For the promotional spikes, they launch additional Spot Instances in an Auto Scaling group, handling the extra load at 70% discount. They set up lifecycle hooks to save session data before Spot termination. This hybrid approach minimizes cost while maintaining performance. Misconfiguration risk: If they had used only On-Demand, costs would be 40% higher; if they had used only Spot, the site could go down during a flash sale when Spot capacity is reclaimed.
Scenario 2: Big Data Analytics Firm
A data analytics company runs nightly batch jobs that process terabytes of log data. The jobs are fault-tolerant and can be restarted. They use a Spot Fleet with multiple instance types (e.g., m5.large, c5.xlarge) across three Availability Zones. They set a maximum bid equal to the On-Demand price and let AWS manage the fleet. They save 80% compared to On-Demand. They also purchase a Compute Savings Plan covering 50% of their expected compute usage to get discounts on both Spot and On-Demand usage. This combination yields the lowest possible cost. What goes wrong: If they misconfigure the Spot Fleet to use only one instance type in one AZ, a capacity shortage could terminate all instances simultaneously, halting the job. They should diversify instance types and AZs.
Scenario 3: Regulated Financial Institution
A bank must run a core banking application on dedicated hardware due to compliance requirements. They use Dedicated Hosts to meet licensing terms for a proprietary database. They purchase 3-year All Upfront Dedicated Hosts to maximize savings. However, they over-provision by 2 hosts, wasting thousands of dollars per month. Solution: They use AWS Config to track host utilization and downsize to the exact number needed. Exam lesson: Dedicated Hosts are expensive; only use when absolutely necessary for licensing or compliance.
What CLF-C02 Tests on This Objective
Objective 4.1: 'Compare AWS pricing models.' The exam focuses on your ability to distinguish between On-Demand, Reserved Instances, Savings Plans, Spot Instances, and Dedicated Hosts. You must know: (1) The discount range for each model (e.g., up to 72% for RIs, up to 90% for Spot). (2) The commitment required (none for On-Demand, 1-3 years for RIs/SPs). (3) The risk of interruption (only for Spot). (4) The flexibility (Savings Plans > RIs). (5) The use cases for each.
Common Wrong Answers and Why Candidates Choose Them
'Reserved Instances are a physical server reservation.' WRONG. RIs are billing discounts; you still launch instances normally. Candidates confuse 'reserved' with 'dedicated'.
'Spot Instances are cheaper than On-Demand but cannot be terminated.' WRONG. Spot can be terminated with 2-minute notice. Candidates think 'spot' is like a fixed-price discount.
'Savings Plans only apply to EC2.' WRONG. Compute Savings Plans also cover Lambda and Fargate. Candidates focus on EC2 because RIs are EC2-specific.
'Dedicated Instances and Dedicated Hosts are the same.' WRONG. Dedicated Hosts give you physical server control; Dedicated Instances do not. Candidates don't read the fine print.
Specific Terms That Appear Verbatim
'Up to 72% discount' for RIs
'Up to 90% discount' for Spot
'1-year or 3-year term' for RIs/SPs
'Compute Savings Plans' vs 'EC2 Instance Savings Plans'
'All Upfront / Partial Upfront / No Upfront' payment options
'2-minute termination notice' for Spot
Tricky Distinctions
Reserved vs Savings Plans: Both require commitment. RIs lock you to a specific instance type; SPs give flexibility across compute services. Exam may ask: 'Which model allows you to change from EC2 to Lambda?' Answer: Compute Savings Plan.
Standard vs Convertible RIs: Convertible allows changes but offers less discount.
Regional vs Zonal RIs: Regional RIs apply to any AZ in a region; Zonal RIs apply to a specific AZ (useful for capacity reservation).
Decision Rule for Multiple Choice
If the question asks for the 'most cost-effective' model for a steady-state workload, look for 'Reserved Instances' or 'Savings Plans'. If the workload is fault-tolerant and flexible, choose 'Spot Instances'. If the workload is unpredictable, choose 'On-Demand'. If the question mentions 'licensing', choose 'Dedicated Hosts'.
On-Demand is pay-as-you-go with no commitment; most expensive per hour.
Reserved Instances offer up to 72% discount for 1- or 3-year commitment to a specific instance type.
Savings Plans offer up to 66% discount (Compute SP) for a commitment to a dollar amount of compute per hour.
Spot Instances offer up to 90% discount but can be terminated with 2-minute notice.
Dedicated Hosts provide a physical server for licensing; Dedicated Instances do not.
Per-second billing applies to Linux instances (minimum 60 seconds).
Free Tier includes 750 hours/month of t2.micro or t3.micro for 12 months.
These come up on the exam all the time. Here's how to tell them apart.
Reserved Instances (RIs)
Commit to specific instance type (e.g., m5.large) in a region or AZ.
Discount up to 72% (Standard) or up to 54% (Convertible).
Only applies to EC2 instances.
Can be shared across accounts via AWS Organizations.
Offering classes: Standard and Convertible.
Savings Plans (SPs)
Commit to dollar amount of compute spend per hour (e.g., $10/hr).
Discount up to 66% (Compute SP) or up to 72% (EC2 Instance SP).
Applies to EC2, Lambda, and Fargate (Compute SP).
Can be shared across accounts via AWS Organizations.
Two types: Compute SP (most flexible) and EC2 Instance SP (less flexible).
On-Demand
No commitment, pay per hour/second.
Highest cost per unit.
Instances run until you stop or terminate them.
Ideal for unpredictable workloads.
No risk of interruption.
Spot Instances
Bid on spare capacity, pay Spot price (discount up to 90%).
Lowest cost per unit.
Instances can be terminated with 2-minute notice.
Ideal for fault-tolerant, stateless workloads.
Interruption risk; not for databases or critical apps.
Mistake
Reserved Instances are a separate type of instance you launch.
Correct
RIs are a billing discount applied to existing On-Demand instances. You launch instances normally; the RI discount is automatically applied to matching usage.
Mistake
Spot Instances can be used for any workload with no risk.
Correct
Spot Instances can be terminated with a 2-minute notice if AWS needs capacity. They are only suitable for fault-tolerant, stateless workloads.
Mistake
Savings Plans only cover EC2 instances.
Correct
Compute Savings Plans also cover AWS Lambda and AWS Fargate usage. EC2 Instance Savings Plans are specific to EC2 instance families.
Mistake
You must pay upfront for Reserved Instances.
Correct
There are three payment options: All Upfront, Partial Upfront, and No Upfront. No Upfront still gives a discount (smaller) with zero upfront payment.
Mistake
Dedicated Instances and Dedicated Hosts are the same thing.
Correct
Dedicated Instances run on hardware dedicated to your account but may share with other instances from your account. Dedicated Hosts give you a physical server with control over instance placement, useful for licensing.
A Reserved Instance (RI) commits you to a specific instance type (e.g., t3.medium) in a specific region for 1 or 3 years. A Savings Plan (SP) commits you to a dollar amount of compute spend per hour (e.g., $10/hour) for 1 or 3 years. SPs are more flexible because they apply to any EC2 instance, Lambda, or Fargate usage (Compute SP). RIs offer slightly higher discounts but lock you into a specific instance family. For the exam, remember that SPs are the newer, more flexible model.
No, Spot Instances are not recommended for databases because they can be terminated with a 2-minute notice. Databases require persistent storage and high availability. Use On-Demand or Reserved Instances for databases. Spot Instances are best for stateless, fault-tolerant workloads like batch processing or CI/CD.
AWS sends a 2-minute termination notice via the instance metadata and CloudWatch Events. You can use this time to checkpoint data, save state, or gracefully shut down. After 2 minutes, the instance is stopped or terminated (depending on your request type). You are not charged for partial hours if terminated by AWS.
No, you can choose No Upfront, Partial Upfront, or All Upfront. No Upfront requires no payment at purchase but has a monthly recurring charge over the term. All Upfront gives the largest discount. For the exam, know the three options and that All Upfront provides the highest savings.
Yes, if your accounts are consolidated under AWS Organizations and you enable RI sharing. The discount is applied to matching instances in any account in the organization. This helps centralize cost management. Exam tip: This is a common feature tested in billing questions.
The AWS Free Tier includes three types: Always Free (e.g., 1 million Lambda requests/month), 12-month free (e.g., 750 hours of t2.micro or t3.micro EC2 per month, 5 GB S3 storage), and short-term trials (e.g., 2 months free for SageMaker). Exceeding these limits incurs standard charges. Exam questions often ask about the 750-hour EC2 limit.
Per-second billing charges for EC2 instances by the second (minimum 60 seconds) for Linux instances launched after 2017. Windows instances are still billed per hour. This reduces costs for short-lived instances. It applies to On-Demand, Reserved, and Spot Instances. Exam tip: Know that per-second billing is for Linux only.
You've just covered AWS Pricing Models — now see how well it sticks with free CLF-C02 practice questions. Full explanations included, no account needed.
Done with this chapter?