The relatively high cost of cloud computing resources, compared to on-premise solutions, is a major challenge for organizations migrating to public cloud services. In this post, we will review several available plans for long-term cost saving of compute resources.

Background

The Pay-As-You-Go, or Pay on-Demand, is the most common option for paying for actual usage when consuming cloud resources. This method is suitable when the required compute power is changing or unpredictable. A good example of this is for services migrated from on-premise to the public cloud (Lift & Shift), or new environments (Dev/Test), and more.

The second most common pricing option is called Spot (Amazon EC2 Spot Instances, Azure Spot Virtual Machines or Google Preemptible Virtual Machines). These options can potentially deliver a discount of up to 90% and are best when there is a demand for large amounts of compute power, and the service is not sensitive to disruptions. Spot Instances are suitable for scenarios when compute power is required by another paying customer and the cloud provider claims the machine back, with 30-second to 2-minute notifications. This method is suitable for image/video processing, batch processing, HPC services, etc.

Reserved Instance

This is the most common pricing option for saving costs. Users commit to one to three years of usage in advance, with a potential savings of up to 70%.

Reserved Instances are available with various payment methods. These range from

  • “All Upfront” – where you pay the entire server cost in advance for the entire commitment period
  • “Partial Upfront” – where you pay the server costs on monthly basis in installments, until the end of the commitment period
  • “No Upfront” – where you pay a fixed price for the server cost until till the end of the commitment period

There are also options for more flexibility on Reserved Instance options. These include Standard RI, where you commit to a certain instance type (instance family type, operating system, payment method, etc.), and Convertible RI, where you are allowed to change the instance type (instance family type, operating system, etc.) during the commitment period.

Additional information about Reserved Instance options can be found at:

It is important to note that these cost-saving options are not limited to virtual servers. It is possible to purchase a commitment for managed services, such as Amazon RDS Reserved Instances, Azure SQL Database reserved capacity, Azure Blob storage reserved capacity, and more.

AWS Saving Plans

AWS has a flexible pricing option, like AWS Reserved Instances, which allows up to 72% discount.

These plans include two alternatives:

  • Compute Saving Plans – This plan allows you to commit to resource consumption in advance, with the flexibility to choose, and change, instance family type, instance size (ratio between CPU/Memory), region, availability zone and operating systems. The Compute Saving plan covers compute resources from virtual machines (EC2 instances), through AWS Fargate and up to Amazon Lambda.
  • EC2 Instance Saving Plans – This plan allows you to save on virtual servers’ costs. However, it is limited to virtual servers from specific instance family types, in a specific region. It is still possible to change instance size (ratio between CPU/Memory), availability zone and operating systems.

Additional information can be found at: https://aws.amazon.com/savingsplans/faq/

Google Sustained Use Discounts

This plan is designed to encourage customers to commit to long term use of Google compute resources, such as virtual servers or Google Kubernetes Engine, for any constant time, longer than 25% of the month. This plan grants an automatic discount of between 20% and 30% of the price list. No action needs to be taken; the discount is applied when reaching the plan’s minimum consumption level of compute resources.

Additional information can be found at: https://cloud.google.com/compute/docs/sustained-use-discounts

Conclusion

The first step toward enjoying long-term cost savings is understanding your compute demands. Studying up and staying up to date on vendors’ various pricing plans and options, then matching those to your needs and environments, is the key to achieving the most cost-effective public cloud solution.


About the author

Eyal Estrin is a cloud architect, working in the Inter-University Computation Center (IUCC) in Israel. He has more than 20 years of experience in infrastructure, information security and public cloud services. He is a public columnist and shares knowledge about cloud services. You can follow him on Twitter at @eyalestrin