A complete overview of EC2 pricing | Estimation methods and cost optimization criteria

It is often thought that EC2 fees are "only for the time the instance is running," but in reality, the total amount is determined by the addition of peripheral costs such as storage (EBS), communication volume, Elastic IP, monitoring, etc. If you start using it without sorting out the system, there are many cases where the estimate and billing differ, resulting in unexpectedly high costs.

In this article, we will organize the billing elements of EC2 fees, differences in payment methods, estimated monthly fees, estimation procedures using the Pricing Calculator, and operational decisions to reduce costs from a practical perspective.

What are EC2 fees charged for?

It is often thought that EC2 charges are "based on the amount of server you start," but in reality, the total amount is determined by a combination of multiple billing elements. When making estimates and operational decisions, it is important to break down and understand the costs by item.

Instance fee (operating time x unit price)

The main cost of EC2 is the usage fee for the virtual server (instance) itself. The fee is basically charged according to the "running time."

• You only pay while the instance is running

• If you stop it, the instance fee will stop.

• Unit price varies greatly depending on the instance type (CPU/memory)

The monthly fee is determined by the number of hours you run the system, so the cost will vary depending on whether you run it 24 hours a day or only when you need it.

Storage fee (EBS capacity/performance)

EC2 often uses storage called EBS (Elastic Block Store), and the cost of this storage is incurred separately from the instance.

The key points are as follows:

• The more GB you have, the higher the price

• Choosing a high-performance volume increases the unit price

• EBS charges remain even if the instance is stopped

If you think that "the charges will stop because the server has been stopped," costs may pile up on the storage side.

Data transfer charges (mainly for outbound communications)

Communication costs are also an often overlooked aspect of EC2 pricing. Outbound communication to the Internet is what is most likely to be charged.

• Web services communicate data back to users

• Integration with external APIs and other companies' systems

• When to transfer backups to an external location

On the other hand, communications within AWS may be free or low-cost, so the design of communication routes has a direct impact on costs.

Additional costs incurred

Additional costs will be incurred if you combine peripheral functions rather than using EC2 alone.

Representative examples are as follows.

• Elastic IP: If you assign a fixed IP, you will be charged according to the conditions.

• Load Balancer: Load balancing usage fee

• CloudWatch: Monitoring and log management costs

These are essential for operation, but if you think that you will only need to pay for the EC2 itself, you may end up with unexpected charges.

Factors that affect EC2 pricing

EC2 fees vary not only depending on billing items but also on the conditions of use. To improve the accuracy of your estimate, you need to understand the main variables that affect the fees.

Pricing varies by instance type

EC2 has different CPU and memory configurations depending on the instance type, and the unit price also changes.

The typical classifications are as follows:

• t-series (burst type): Suitable for small-scale web and testing, relatively cheap

• m series (general-purpose): The standard type most commonly used in business systems

• C series (calculation optimization): Suitable for processing with high CPU load, and unit price tends to increase

If you choose excessive specifications for your application, it will become a fixed cost. The first step in optimizing your costs is to "select the right size type."

Monthly fee determined by operating hours

EC2 is basically charged for the time it is running. Even with the same configuration, the monthly fee will vary depending on the running time.

• Production environment (24-hour operation): Monthly fee is stable but becomes a fixed cost

• Development and testing environment (start only when necessary): Significant reductions can be achieved by stopping operation

The cost difference is particularly clear when simply shutting down the development environment at night or on holidays.

Price differences by region

EC2 unit prices vary by region. Even for the same instance type, prices vary by region.

• Tokyo region tends to be more expensive than US region

• Varies depending on exchange rates and supply and demand conditions

However, choosing based on price alone can lead to issues with latency and data location requirements, so you need to balance cost and requirements.

OS/license differences (Windows is an additional cost)

EC2 pricing varies depending on the OS type. Windows in particular has a higher unit price per instance than Linux, as the license fee is included in the usage fee.

Linux is available with a relatively simple pricing structure, but Windows has a higher monthly fee even with the same specifications, so it is important to make sure you have the OS requirements in place at the quote stage.

The fact that the monthly fee varies depending on the OS is something you should definitely keep in mind when getting a quote.

Differences in payment methods

EC2 pricing is not determined solely by the "unit price of the instance." Even with the same configuration, monthly costs will vary depending on the payment method you choose. The three most common options are on-demand, reserved/savings plans, and spot.

On-demand: Suitable for short-term testing

On-demand is the most basic method, where you start it when you need it and pay only for what you use. While it is flexible and there are no contractual restrictions, it is important to note that the unit price tends to be higher, so if you run it constantly in production, costs can easily increase.

Reserved/Savings Plans: The basis for long-term management

Reserved Instances (RIs) and Savings Plans are systems that allow you to receive discounts on a one- or three-year contract. They are most effective in production environments that run continuously for long periods of time, and are central to optimizing EC2 costs.

Spot: Effective for batch and one-time processing

Spot instances are a way to use unused AWS capacity at low cost. However, because they may be interrupted depending on demand, they should only be used for "applications that can function even if they are stopped," such as batch processing or temporary verification.

Criteria for selection

Rather than simply choosing the cheapest payment method, it is more practical to use different payment methods depending on the purpose. By using RI/Savings Plans for fixed production operations, combining on-demand with suspended operations for development and testing, and using spot services for peak times and temporary processing, it becomes easier to balance cost and stability.

Get a rough idea of ​​monthly fees

When considering EC2 pricing, the first thing you want to know is, "How much will it cost per month?" However, because EC2 is a pay-as-you-go system, the cost is not a fixed monthly amount, but varies depending on the configuration and operation. Here we will provide a rough idea of ​​the cost using typical patterns.

Small-scale web server example

The simplest case is when running a small website or admin screen. If you run a small instance of t-series all the time and keep storage to a minimum, the monthly cost can be as low as a few thousand yen to 1 yen.

On the other hand, low traffic does not necessarily mean low cost, and costs can add up when adding backups and monitoring. Even on a small scale, peripheral costs will be incurred.

Examples of business systems

For internal business applications or mission-critical applications, it is common to configure multiple general-purpose instances such as m-series instances. Monthly costs tend to be in the tens of thousands to hundreds of thousands of yen, and this increases further if a multi-AZ configuration or load balancer is used for availability.

In a production environment, estimates should include the services required for redundancy and operation rather than just the "server fee."

Cost-sensitive points

The typical reason why EC2 charges are higher than expected is not the unit price of the instance but peripheral factors, with communication and storage being the most affected.

Increasing the amount of outbound data transfer will increase communication costs. Also, since EBS continues to charge even if the instance is stopped, costs will pile up if unnecessary volumes or snapshots remain.

When getting an idea of ​​the monthly cost, it is important not to calculate only by the instance.

Steps to estimate pricing using the AWS Pricing Calculator

EC2 fees vary depending on the configuration and operating conditions, so if you try to guess the approximate monthly cost based on your intuition, you will end up with an inaccurate result. To get a sense of the cost, the basic way to do it is to use the official AWS Pricing Calculator.

Items to enter

The Pricing Calculator allows you to calculate a monthly estimate by entering your EC2 configuration. The four minimum items you need to know when getting an estimate are the instance type, operating hours, EBS (capacity and type), and data transfer amount.

The cost of an instance is determined by the instance type and operating hours, and the monthly cost varies greatly depending on whether it is running 24 hours a day or is stopped. Furthermore, EBS charges remain even if the instance is stopped, so capacity and performance must be taken into consideration.

Another factor that is often overlooked is the amount of data transferred. Outbound communications to the Internet in particular tend to incur costs, so it is important to estimate the amount based on figures that closely reflect actual operations.

Points that are likely to miss estimates

The reason why the estimate and actual billing may differ is because there are "unentered costs." Typical examples are snapshots and logs. If you regularly take backups, storage costs will continue to accumulate.

The usage fees for peripheral services such as NAT Gateway and load balancers are charged separately from the EC2 instance itself. If you intend to include them in your configuration, you will need to include them in your estimate from the beginning.

Tips for calculating by composition

When using the Pricing Calculator, it is more accurate to consider the components separately rather than "estimating everything in one go."

For example, the uptime differs between a production environment and a development environment, and the optimal payment method differs between a web server and a batch processing environment.

Breaking it down by use and making estimates makes it easier to see where the main costs lie, and makes it easier to make optimization decisions.

Typical patterns of increased charges and points to note

A common problem with EC2 charges is that even if the estimate for the unit price of an instance is correct, peripheral costs accumulate during operation, causing the bill to balloon. Here we will summarize some typical patterns that frequently occur.

EBS charges remain even if the instance is stopped

If you stop an EC2 instance, the instance fee will stop. However, since storage (EBS) is charged separately, the cost will remain even if you stop the instance.

Even if you think you have stopped the environment created for testing, if there is volume remaining, the monthly fee will continue. The most commonly misunderstood point is that "stopping" does not mean "free."

Snapshot log accumulation

Snapshots for backup purposes and monitoring logs will also pile up if left unattended.

Once you start using the system, the amount of "automatically collected data" will continue to increase. Even if it may seem small on its own, if it remains for a long period of time it will become a non-negligible cost. In a production environment, backup operations and deletion rules must be decided from the beginning.

Overlooking peripheral functions such as NAT Gateway

The most common cause of sudden increases in billing is not EC2 itself, but peripheral services. A typical example is a NAT Gateway. This may be required for configurations that allow external communication, but usage fees and forwarding fees will be charged separately.

Similarly, costs for load balancers and monitoring functions will be incurred "only as much as you add them." It is essential to estimate costs for the entire configuration, not just for individual EC2 instances.

Forgetting to stop the development environment

The most common problem is the neglect of development and testing environments. Even if the production environment is managed, the development side tends to leave "temporarily set up instances" that are left behind, and there are cases where you are constantly being charged without realizing it.

Simply creating a system for downtime scheduling in the development environment can lead to significant savings.

Key design and operation points to reduce EC2 costs

EC2 costs cannot be optimized simply by choosing a cheap instance. The key is to create a situation where costs are naturally kept down based on the configuration and operation.

First, get the size right

The first thing to try is to review the instance size. As long as you are running it with excessive specs, there will be waste no matter what discount plan you use.

By looking at CPU and memory usage, you can change your fixed costs by simply choosing the type that best suits your situation. Cost optimization starts with "not spending more than necessary."

RI/Savings Plans are applied from fixed operation

If you're looking for a discount, a production environment that runs 24/7 is your top priority.

RIs and Savings Plans are designed for long-term use, so their effectiveness is maximized when applied to instances that run stably. Conversely, if you apply them to an environment where usage is uncertain, the loss of flexibility will outweigh the discount.

Increase only peak demand with Auto Scaling

The advantage of EC2 is that you can adjust the number of units to suit your needs. Instead of always using the maximum configuration, you can reduce waste by increasing the number of units only during peak times.

For services with fluctuating access, a configuration that assumes auto scaling has a major impact on cost efficiency.

Monitor with Budget and Cost Explorer

Cost optimization is not something that is done once and then finished. During operation, unnecessary resources may increase, and communication volume may increase unexpectedly.

Simply notifying you of budget overruns with AWS Budgets and visualizing the causes of increases with Cost Explorer will make it easier to prevent runaway charges. It is important to have a system that can detect issues early in your operations, rather than just noticing after seeing your bill.

Conclusion

EC2 fees are not determined solely by the amount of time the instance is running. The total price includes storage (EBS), communication volume, and peripheral services.

To properly manage fees, it is important to first break down the billing elements, sort out the differences due to instance type, operating time, and payment method, and then use the Pricing Calculator to estimate the fees for each configuration.

In addition, the actual causes of inflated bills are primarily operational issues such as residual EBS, accumulation of snapshots, overlooking peripheral functions such as NAT Gateway, and neglecting development environments.

EC2 costs can be adequately controlled by understanding how it works and by developing good design and monitoring habits.

Do you have any concerns about AWS?

If you have any questions or concerns about using AWS, estimates, configuration, operation, etc., please feel free to contact us. We will help you make a smooth decision by establishing a common understanding with the local team and clarifying prerequisites.

This service, "IIJ Managed Cloud for AWS," is jointly provided by the IIJ Group, Japan's first commercial Internet service provider, and Serverworks, an AWS Premier Tier Service Partner. It is compatible with global environments, including Southeast Asia, and provides AWS support tailored to on-site decisions.

▶ Check out the detailed documentation
▶ Consult with us about using AWS