Skip to content
Draft: this topic has not been reviewed yet. Check facts against the linked AWS documentation before relying on them.

Cloud economics — why the cloud can cost less

Cloud Practitioner · Domain 1 · Cloud ConceptsTask 1.4: Understand concepts of cloud economicsLevel 116 minDraft
Difficulty
Confidence

The cloud changes how you pay for IT, not just how much. To compare honestly, you have to count every cost of running your own servers, then look at the ways the cloud lets you spend less.

Fixed costs vs variable costs

On premises On AWS
How you pay Large purchases up front (capital expense, often called CapEx) Monthly bills for what you used (operating expense, OpEx)
When usage drops You still own, power and maintain the hardware Your bill drops
Starting a project Wait for budget approval and delivery of hardware Start today with a small amount
Risk if the project fails Hardware you no longer need You stop the resources and stop paying

Turning fixed costs into variable costs is the single biggest economic change. It lowers the cost of trying new ideas and removes the risk of buying the wrong amount.

The full cost of running on premises

Comparing the price of a server with an hourly cloud price is misleading. The total cost of ownership (TCO) of your own data center includes:

  • Hardware: servers, storage, network switches, plus spares, replaced every few years.
  • Facilities: floor space, power, cooling, fire suppression and physical security.
  • Software licenses for operating systems, databases and virtualization.
  • People: staff to buy, install, patch, monitor and repair everything, often around the clock.
  • Idle capacity: everything bought “just in case” for peaks or failures, which mostly sits unused.

Many of these costs are invisible in a simple price comparison, but they are real. In the cloud, most of them are included in the price you pay AWS.

Licensing: bring your own or included

Commercial software, such as Microsoft Windows Server or Oracle Database, needs licenses. On AWS you have two ways to cover them.

Model How it works When it fits
License included The license is part of the AWS price, e.g. Windows Server on EC2 or some Amazon RDS database engines You have no licenses to reuse, or want simple pay-as-you-go
Bring Your Own License (BYOL) You reuse licenses you already own You have already paid for licenses. Some licenses are tied to physical cores or sockets and need a Dedicated Host

AWS License Manager tracks how many licenses you are using, on AWS and on premises, so you stay within your agreements.

Rightsizing: pay for what the workload really needs

Rightsizing means matching an instance’s type and size to what the workload actually uses. Servers sized for a guess are often mostly idle: a server running at 10% CPU all month is costing you for the other 90%.

  • AWS Compute Optimizer analyzes usage and recommends better sizes for EC2 instances and other resources.
  • Cost Explorer also shows rightsizing recommendations.
  • Because cloud resources can be resized quickly, you can start small and adjust, which is impossible with bought hardware.

Automation saves money too

  • Infrastructure as code builds environments the same way every time, so there are fewer costly mistakes and less manual work.
  • Schedules can stop development and test environments at night and on weekends, when nobody uses them.
  • Auto Scaling adds capacity only while demand needs it, then removes it.
  • Managed services handle patching, backups and failover, so fewer staff hours go into routine tasks.

Economies of scale

AWS buys enormous amounts of hardware, power and bandwidth. That scale lowers its cost per unit, and AWS passes savings on through lower prices for its services. One company running its own data center cannot reach the same purchasing power.

Putting it together: a reporting system

A company runs a reporting system on four large servers it bought three years ago. They are busy for two days at month end and nearly idle the rest of the time.

  1. Moving to AWS removes the next hardware refresh and the data center costs.
  2. Rightsizing shows that smaller instances handle normal days easily.
  3. Auto Scaling adds capacity for the two busy days.
  4. A schedule turns the test environment off outside office hours.

The company pays for heavy capacity only two days a month instead of owning it all year.

Common mix-ups

  • Rightsizing vs pricing models. Rightsizing is choosing the right size. Choosing Spot Instances or Savings Plans is a pricing decision. Both save money in different ways.
  • CapEx vs OpEx. Buying servers is capital expense; paying a monthly AWS bill is operating expense.
  • License included is not always cheaper. If you already own licenses, BYOL may cost less.

Check yourself

A company’s on-premises servers run at about 15% CPU on average. What cloud practice would cut costs most directly?

Rightsizing. Move to smaller instance types that match the real usage; AWS Compute Optimizer can recommend them.

Which on-premises costs are often forgotten when comparing with cloud prices?

Power, cooling, floor space, physical security, hardware refresh, spare capacity, and the staff time to install, patch and repair servers.

A company already owns Windows Server licenses that are tied to physical cores. How can it use them on AWS?

With Bring Your Own License (BYOL), usually on Dedicated Hosts, which give it a physical server so per-core licensing rules can be met.

Key takeaways

  • The cloud turns large fixed costs into variable costs you can stop.
  • Compare total cost of ownership, including power, space, staff and idle capacity.
  • Save more by rightsizing, automating schedules and scaling, and choosing the right licensing model.

Cloud Study Lab is an independent study site. It is not affiliated with or endorsed by Amazon Web Services. AWS and the names of AWS certifications are trademarks of Amazon.com, Inc. or its affiliates.