Reviewing your cost structure

AWS provides a number of options for you to balance the need for instance flexibility and cost savings. This section summarizes common recommendations to help you gain visibility and manage cost. For a deep dive on cost optimization best practices, see the links provided in each of the subsections.

Cost metrics and workload management

Talk to your finance team about the need to run small test fleets during the testing and benchmarking stage. Look at flexible pricing options such as Spot Instances to reduce the cost of research and development. For more information, see the Leveraging Amazon EC2 Spot Instances at Scale white paper. Develop specific cost-related metrics that allow you to measure resource utilization, instances turned off daily, and instances to which cost tags have been applied. Other measures include workload management and the ability to turn off development, test, and staging workloads when not in use. This enables you to keep cost low while enabling development teams to perform test and benchmarking tasks on multiple instance types and architectures.

Instance fleets

Phasing in AWS Graviton-backed instances into your production fleets may require you to run multiple instance types and support different architectures. Consider optimizing your production fleets for instance type flexibility by making use of constructs such as Amazon EC2 Fleets and Compute Savings Plans. With Amazon EC2 Fleet, you provision capacity across EC2 instance types and across purchase models to achieve your desired scale, performance and cost. You can combine On-Demand and Spot purchasing options and specify an unlimited number of instance types. With Amazon EC2 Compute Savings Plans, you gain the greatest flexibility in terms of instance family, size, Availability Zone (AZ), region, operating system or tenancy at an up to 66% lower cost compared to On-Demand rates. Use Compute Savings Plans to maximize cost savings across parameters such as instance type, architecture, operating system, tenancy model, and region. Additional capabilities include multi-architecture support in Auto-Scaling Groups (ASG), which enables you to configure both Graviton2 and x86-based Amazon EC2 Instances in the same Auto-Scaling group with different AMIs.

Pricing model

Select the right pricing model for your workload as you ramp up your AWS Graviton usage. Amazon EC2 On Demand allows you to pay for compute capacity by the hour or the second (depending on which instances you run) without longer-term commitments or upfront payments. Amazon EC2 Spot Instances allow you to request spare Amazon EC2 computing capacity to run fault-tolerant (interruptible) workloads for up to 90% off the On-Demand price. Amazon EC2 Reserved Instances provide deep discounts (up to 72% lower compared to On Demand prices) when you commit to a 1- or 3-year term and instance type. Finally, Amazon Savings Plans offer significant savings over On Demand, just like EC2 Reserved Instances, in exchange for a commitment to use a specific amount of compute power. Compute Savings Plans automatically apply to EC2 instance usage regardless of instance family, size, AZ, region, OS, or tenancy.