Amazon EC2 Pricing Options and Billing Decisions

Purpose of This Lesson

This lesson explains how to select and operate Amazon EC2 pricing options based on workload predictability, cost objectives, capacity requirements, hardware isolation, and licensing constraints. It also covers EC2 billing granularity, Reserved Instance modification and exchange capabilities, Savings Plans, Spot interruption behavior, and the differences between Dedicated Instances and Dedicated Hosts.

Because this material is part of AWS Certified CloudOps Engineer Associate preparation, focus on recognizing the workload scenario and selecting the pricing or capacity feature that best fits it.

Key Concepts

• On-Demand Instances: Standard pricing with no long-term commitment. They are suitable for development and testing, short-term workloads, and workloads whose demand is unpredictable.
• Reserved Instances (RIs): A one- or three-year commitment that can provide a substantial discount, described in the lesson as up to 75%. RIs are appropriate for steady-state, predictable workloads.
• Spot Instances: Use unused EC2 capacity at potentially large discounts, described as up to 90%. AWS can reclaim the capacity, so the workload must tolerate interruption.
• Dedicated Instances: Instances running on hardware physically isolated from instances belonging to other AWS customers. Billing is per instance.
• Dedicated Hosts: Entire physical servers dedicated to one customer. Billing is per host, with visibility into sockets, cores, and host IDs. This can support server-bound software licensing and controlled host placement.
• Savings Plans: A one- or three-year commitment to a consistent amount of usage, expressed as a dollar-per-hour commitment. Savings Plans can apply to compute services such as EC2, Fargate, and Lambda, depending on the plan type.
• On-Demand Capacity Reservations: Reserve EC2 capacity in a specific Availability Zone for any duration without a long-term term commitment. They address capacity availability, not primarily discounting.

EC2 Pricing and Capacity Models

On-Demand Pricing

On-Demand pricing has no discount and requires no commitment. Choose it when:

• Instances are used for development or testing.
• The workload is short-lived.
• Demand is unpredictable.
• Committing to a one- or three-year term would create unnecessary risk.

On-Demand is also the baseline against which discounted options are commonly compared.

Reserved Instances

Reserved Instances are intended for workloads with stable, predictable usage. The commitment is for one or three years, in exchange for a discount.

The lesson distinguishes two types:

• Standard Reserved Instances: Permit changes to the Availability Zone, Linux instance size, and networking type through the Modify Reserved Instances API.
• Convertible Reserved Instances: Permit the Standard RI changes and also allow changes to the instance family, operating system, tenancy, and payment option through the Exchange Reserved Instances API.

Payment choices are:

• All upfront: Provides the largest discount among the payment options covered.
• Partial upfront: Combines an upfront payment with ongoing payments.
• No upfront: Uses ongoing payments and provides a lower discount than paying fully upfront.

RI discounts apply when the attributes of running usage match the attributes of the Reserved Instance. Relevant attributes include instance type, operating system, tenancy, Availability Zone, and Region.

A Regional reservation and an AZ-specific reservation have different capacity implications:

• Reserving in a specific Availability Zone can reserve capacity in that AZ.
• Reserving by Region applies the discount across Availability Zones in that Region, but does not reserve capacity.

Savings Plans

Savings Plans use an hourly spending commitment rather than tying the commitment only to one exact instance configuration.

• Compute Savings Plans: One- or three-year commitment to a consistent hourly amount of compute usage. They can apply across EC2, Fargate, and Lambda, including changes across Region, instance family, size, tenancy, and operating system as described in the lesson.
• EC2 Instance Savings Plans: One- or three-year hourly commitment for EC2 within a selected Region and instance family. Instance size, tenancy, and operating system can vary within that scope.

The central distinction is flexibility: Compute Savings Plans cover a broader set of compute services and changes, while EC2 Instance Savings Plans are more specifically scoped to EC2 in a selected Region and instance family.

Spot Instances and Fleet Options

Spot Instances use unused EC2 capacity at a major discount, but AWS may reclaim that capacity. A reclaimed Spot Instance receives a two-minute warning through instance metadata and CloudWatch events, allowing automation to save state or begin shutdown processing.

The interruption risk makes Spot appropriate for fault-tolerant, interruptible, or flexible workloads—not workloads that require uninterrupted execution on a particular instance.

Fleet capabilities described in the lesson include:

• Spot Fleet: Launches and maintains a target capacity using Spot and On-Demand Instances.
• EC2 Fleet: Uses one API call to launch and maintain specified numbers of Spot, On-Demand, and Reserved Instances. Separate Spot and On-Demand capacity targets, prices, instance types, and Availability Zones can be specified.

Spot Blocks

A Spot Block is intended for workloads that need uninterrupted execution for a defined period of 1–6 hours. You specify the block duration in minutes, instance count, and the price you are willing to pay.

The lesson describes Spot Block pricing as approximately 30–45% below On-Demand pricing. This is a smaller discount than regular Spot pricing, but the tradeoff is a specified uninterrupted run period. Spot Blocks should not be confused with ordinary Spot Instances, which remain subject to capacity reclamation.

EC2 Billing and Capacity Reservations

Billing Granularity

For the operating systems and purchase models identified in the lesson, billing can occur per second with a one-minute minimum:

• Amazon Linux
• Windows
• Ubuntu
• On-Demand, Reserved, and Spot forms of those instances
• EBS volumes

Commercial Linux distributions such as Red Hat Enterprise Linux and SUSE Enterprise Server are billed per hour with a one-hour minimum, according to the lesson.

Always distinguish the billing unit from the pricing model. For example, an instance can use Spot pricing while still being subject to the applicable per-second billing rules and minimum charge described above.

On-Demand Capacity Reservations

An On-Demand Capacity Reservation reserves compute capacity for EC2 instances in a selected Availability Zone. It does not require a one- or three-year commitment and can be canceled at any time.

When creating one, specify:

• Availability Zone
• Number of instances
• Instance type
• Tenancy
• Platform or operating system

Use a Capacity Reservation when the primary concern is ensuring that On-Demand capacity will be available in a particular AZ. It is different from an RI or Savings Plan: those primarily address discounted pricing, while a Capacity Reservation addresses capacity availability.

Exam- or Assessment-Relevant Takeaways

• Match On-Demand to short-term, development/test, or unpredictable usage.
• Match Reserved Instances to predictable, steady-state EC2 usage requiring a term commitment.
• Remember that an AZ-specific RI can reserve capacity, while a Regional RI applies discounts across AZs without reserving capacity.
• Use Standard RIs when limited configuration changes are sufficient; use Convertible RIs when changes to family, OS, tenancy, or payment option may be required.
• Select Compute Savings Plans for broader compute flexibility across EC2, Fargate, and Lambda.
• Select EC2 Instance Savings Plans when the commitment is specifically EC2 within a chosen Region and instance family.
• Use Spot Instances only when the workload can tolerate interruption and respond to the two-minute warning.
• Use a Spot Fleet or EC2 Fleet when capacity should be managed across multiple instance purchase options or configurations.
• Choose a Spot Block when a 1–6 hour uninterrupted period is required and a smaller discount is acceptable.
• Choose Dedicated Hosts when socket/core visibility, host affinity, targeted placement, or server-bound licensing is important.
• Do not confuse physical isolation with host-level control: Dedicated Instances provide isolation, while Dedicated Hosts provide an entire physical server and additional hardware visibility.
• Use an On-Demand Capacity Reservation when the key requirement is guaranteed capacity in a specific AZ without a long-term commitment.

Tool / Feature Decision Guide

Requirement	Best-fit option	Decisive reason
Short-term or unpredictable workload	On-Demand	No commitment is required.
Stable EC2 usage for one or three years	Reserved Instance	Provides a term-based discount for predictable usage.
Stable usage but broader compute-service flexibility	Compute Savings Plan	Applies to EC2, Fargate, and Lambda within the commitment model.
EC2 usage in one Region and instance family with size flexibility	EC2 Instance Savings Plan	Keeps the commitment within a selected EC2 Region and family.
Interruptible batch or fault-tolerant processing	Spot Instance	Low cost, with the risk of reclamation.
Mixed Spot and On-Demand target capacity	Spot Fleet or EC2 Fleet	Fleet tools maintain specified capacity using multiple purchase options.
One to six hours of uninterrupted discounted execution	Spot Block	Provides a specified block duration at a smaller discount than regular Spot.
Capacity must be available in a specific AZ	On-Demand Capacity Reservation	Reserves capacity without a long-term term commitment.
Physical isolation only, billed per instance	Dedicated Instance	Isolates instances from other customers at the host hardware level.
Full physical server, socket/core visibility, or license binding	Dedicated Host	Provides host-level control and is billed per host.

Common Traps / Misconceptions

• A discount does not automatically mean capacity is reserved. Regional Reserved Instances provide discount applicability across AZs but do not reserve capacity. Capacity Reservations address availability in a specific AZ.
• Spot is not simply a cheaper On-Demand option. AWS can reclaim Spot capacity, so interruption handling is a design requirement.
• Spot Block is not the same as ordinary Spot. A Spot Block provides a defined uninterrupted period in the scenario described, while regular Spot remains reclaimable.
• Dedicated Instances and Dedicated Hosts are not interchangeable. Dedicated Instances are billed per instance; Dedicated Hosts are billed per host and provide hardware-level visibility and placement controls.
• Standard and Convertible RIs have different change capabilities. Convertible RIs permit broader changes, including family, OS, tenancy, and payment option.
• An RI and a Capacity Reservation solve different problems. An RI is primarily a pricing commitment; a Capacity Reservation is primarily a capacity-availability mechanism.
• Per-second billing does not mean there is no minimum charge. The lesson specifies a one-minute minimum for the applicable systems and resources.
• Savings Plans are not all equivalent. Compute Savings Plans have broader service and configuration flexibility than EC2 Instance Savings Plans.

Real-World Engineer / Analyst Notes

• Treat Spot interruption handling as part of the application design. Use the two-minute warning delivered through instance metadata and CloudWatch events to save state, drain work, or replace capacity.
• Review actual usage before selecting a long-term commitment. A discount can become costly if workload demand or configuration changes make the commitment difficult to use.
• Separate cost optimization from capacity assurance. A team may need both a discounted pricing commitment and an AZ-specific Capacity Reservation, depending on the operational requirement.
• Dedicated Hosts can be justified by licensing and placement requirements, but they are more expensive because billing is per physical host rather than per instance.
• For clustered workloads, host affinity and targeted placement can help ensure that instances run on separate Dedicated Hosts, reducing correlation with a single hardware failure.
• Validate current AWS pricing and service documentation before implementation. The discount percentages and service details in this lesson are study context and may change over time.

Quick Reference Summary

• On-Demand: Flexible, no commitment, no discount.
• Reserved Instance: One- or three-year EC2 commitment for predictable usage; Standard has fewer modification options than Convertible.
• Compute Savings Plan: One- or three-year hourly commitment spanning EC2, Fargate, and Lambda.
• EC2 Instance Savings Plan: One- or three-year hourly commitment for EC2 in a selected Region and instance family.
• Spot: Lowest-cost flexible capacity, but AWS may reclaim it with a two-minute warning.
• Spot Fleet: Maintains target capacity using Spot and On-Demand Instances.
• EC2 Fleet: Manages Spot, On-Demand, and Reserved capacity through one API call.
• Spot Block: Defined 1–6 hour uninterrupted period at a smaller discount than regular Spot.
• Dedicated Instance: Hardware isolation, billed per instance.
• Dedicated Host: Entire physical server, socket/core visibility and host affinity, billed per host.
• Capacity Reservation: Reserves On-Demand capacity in a specific AZ with no term commitment.
• Billing: Per-second with a one-minute minimum for the listed systems and EBS; commercial Linux distributions in the lesson use hourly billing with a one-hour minimum.

Flashcards

Q: A development environment has unpredictable usage and may be deleted after short testing sessions. Which EC2 pricing option is the safest default and why?
A: On-Demand pricing, because it requires no long-term commitment and is suited to short-term or unpredictable workloads.

Q: When is a Reserved Instance a better fit than On-Demand pricing?
A: When EC2 usage is steady and predictable for a one- or three-year period, making the commitment worthwhile in exchange for a discount.

Q: What is the key difference between a Regional RI and an AZ-specific RI?
A: An AZ-specific RI can reserve capacity in that AZ. A Regional RI applies its discount across AZs in the Region but does not reserve capacity.

Q: A team expects to change instance family, operating system, and tenancy during its commitment. Which RI type should it evaluate?
A: A Convertible Reserved Instance, because it supports those broader changes through the Exchange Reserved Instances API.

Q: When would you choose a Compute Savings Plan instead of an EC2 Instance Savings Plan?
A: Choose a Compute Savings Plan when the commitment may span EC2, Fargate, or Lambda and needs broader flexibility. An EC2 Instance Savings Plan is scoped to EC2 in a selected Region and instance family.

Q: What is the main operational risk of Spot Instances?
A: AWS can reclaim the capacity. The workload must tolerate interruption and should use the two-minute warning from instance metadata and CloudWatch events to respond.

Q: A batch system can use either Spot or On-Demand capacity and must maintain separate target amounts for each. Which feature fits this requirement?
A: A Spot Fleet or EC2 Fleet can maintain target capacity using multiple purchase options; the lesson specifically describes separate Spot and On-Demand targets for fleet configurations.

Q: When is a Spot Block preferable to an ordinary Spot Instance?
A: When the workload needs a defined uninterrupted execution period of 1–6 hours and accepts a smaller discount than regular Spot pricing.

Q: Which option provides physical isolation but is billed per instance rather than per host?
A: A Dedicated Instance provides physical isolation from other customers’ instances and is billed per instance.

Q: Why would an organization select a Dedicated Host?
A: It provides an entire physical server with socket/core visibility, host ID visibility, affinity, and targeted placement. These capabilities can help with server-bound software licenses or host-aware clustering.

Q: What is the difference between modifying a Standard RI and exchanging a Convertible RI?
A: A Standard RI uses the Modify Reserved Instances API for a narrower set of changes. A Convertible RI uses the Exchange Reserved Instances API and supports additional changes such as family, OS, tenancy, and payment option.

Q: An operations team must ensure that EC2 capacity is available in one AZ but does not want a one- or three-year commitment. What should it use?
A: An On-Demand Capacity Reservation, which reserves capacity in a specified AZ for a chosen duration and can be canceled at any time.

Q: Which operating systems in the lesson use per-second billing with a one-minute minimum for the listed EC2 purchase models?
A: Amazon Linux, Windows, and Ubuntu. The lesson also states that EBS volumes use per-second billing with a one-minute minimum.

Q: What billing rule does the lesson give for Red Hat Enterprise Linux and SUSE Enterprise Server?
A: They are billed per hour with a one-hour minimum, according to the lesson.

Practice Questions

Question 1

A company runs a noncritical image-processing workload that can checkpoint progress and restart elsewhere. It wants the lowest possible EC2 cost and can respond to interruption notices. Which option is the best fit?

A. Dedicated Host
B. Spot Instances
C. Standard Reserved Instances
D. On-Demand Capacity Reservation

Correct answer: B. Spot Instances

Explanation: The workload is interruptible and can checkpoint state, which addresses the central risk of Spot capacity reclamation. The other options either prioritize isolation, long-term predictability, or capacity availability rather than lowest flexible compute cost.

Question 2

A production service has predictable EC2 usage for three years. During that period, the team may change the instance family, operating system, and tenancy. Which pricing option best matches the stated requirements?

A. Convertible Reserved Instance
B. Standard Reserved Instance
C. Spot Block
D. On-Demand pricing

Correct answer: A. Convertible Reserved Instance

Explanation: Convertible RIs support the broader configuration changes listed in the scenario through the Exchange Reserved Instances API and retain the benefit of a term-based commitment.

Question 3

An application must run On-Demand EC2 instances in a particular Availability Zone tomorrow. The team does not want a long-term commitment, but capacity availability is more important than receiving a pricing discount. What should it create?

A. Regional Reserved Instance
B. Compute Savings Plan
C. On-Demand Capacity Reservation
D. Spot Fleet

Correct answer: C. On-Demand Capacity Reservation

Explanation: A Capacity Reservation is designed to reserve compute capacity in a specific AZ without requiring a one- or three-year term. A Regional RI provides discount applicability but does not reserve capacity.

Question 4

A software license is bound to physical CPU sockets, and administrators need to control which physical hosts run instances in a cluster. Which option is most appropriate?

A. Dedicated Instance
B. Dedicated Host
C. Spot Instance
D. Regional Reserved Instance

Correct answer: B. Dedicated Host

Explanation: Dedicated Hosts provide socket/core visibility, host ID visibility, affinity, and targeted instance placement. Those capabilities support server-bound licensing and separation of clustered instances across hosts.

Question 5

A platform may run workloads on EC2, Fargate, and Lambda across changing Regions, instance families, sizes, tenancies, and operating systems. It wants a one- or three-year hourly usage commitment. Which plan is the strongest match?

A. EC2 Instance Savings Plan
B. Compute Savings Plan
C. Standard Reserved Instance
D. Spot Block

Correct answer: B. Compute Savings Plan

Explanation: The scenario requires cross-service compute coverage and broad flexibility. The lesson describes Compute Savings Plans as applying to EC2, Fargate, and Lambda with broader configuration flexibility than EC2 Instance Savings Plans.

WordPress Metadata

Suggested Slug:
amazon-ec2-pricing-options-billing-soa-c03

Meta Description:
Study Amazon EC2 pricing models, billing rules, Reserved Instances, Savings Plans, Spot capacity, Dedicated Hosts, and capacity reservations for the AWS Certified CloudOps Engineer Associate exam.

Tags:
Amazon EC2, EC2 pricing, On-Demand Instances, Reserved Instances, Spot Instances, Savings Plans, Dedicated Hosts, Dedicated Instances, Capacity Reservations, EBS billing, AWS CloudOps, SOA-C03