Many organizations rely heavily on cloud infrastructure deployed on AWS, which can eat up a significant percentage of the IT budget. Consider using AWS Savings Plans to help reduce costs.
Cloud cost optimization best practices in AWS include cost budgeting, monitoring and infrastructure optimizations. AWS also offers tools to implement FinOps best practices. AWS Savings Plans can provide significant discounts on compute utilization costs and improve ROI for cloud deployments.
AWS Savings Plans deliver cost reductions by committing to a fixed hourly spend on specific usage types related to AWS compute infrastructure. They support commitments for one or three years, and offer the option to pay monthly or using partial-upfront or all-upfront scenarios. The higher the commitment term and the upfront payment, the higher the cost reductions, which get applied to all eligible resources within an AWS Organization.
Savings Plan options to consider
There are several types of savings plans available:
- Compute Savings Plans. Cover multiple compute services, such as Lambda, EC2 instances and Fargate across multiple AWS regions.
- EC2 Instance Savings Plans. Commitments are specific to an instance family and AWS region.
- SageMaker Savings Plans. Commitments cover all instance families and regions supported by SageMaker AI.
- Database Savings Plans. Commitments apply to all regions, instance families and DB engines supported by Amazon RDS.
A key difference between Compute Savings Plans and EC2 Instance Savings Plans is flexibility. While EC2 Savings Plans only cover a specific instance family, Compute Savings Plans cover multiple instance types and compute resources, such as Lambda functions and containers deployed using Fargate.
However, the cost reductions delivered by Compute Savings Plans can be slightly lower, so evaluate how much flexibility the organization requires. For systems that have been recently launched, it is important to have greater flexibility to apply potential fine-tunings and still receive the cost benefits from Savings Plans.
When planning the IT budget for AWS Savings Plans, CTOs should note that there is a wide range of available savings. For example, EC2 Savings Plans offer up to 72% savings, compared to the 66% savings possible with Compute Savings Plans. SageMaker AI Savings Plans can deliver up to 64% in savings, and Database Savings Plans can deliver up to 35%. When executing the actual cost calculations, clarify the type of compute utilization covered by the selected Savings Plan and whether the organization should pay upfront or monthly.
The cost reduction difference between monthly and all-upfront payments is significant. For example, a no-upfront one-year Compute Savings Plan can deliver approximately 26% savings depending on the usage type, while a three-year all-upfront commitment can deliver up to 66% in savings. Upfront payments yield higher savings, but some businesses are negatively affected by high upfront costs. For example, an upfront three-year $10/hour commitment would result in a one-time payment of close to $263K, while the no-upfront option would result in approximately $7,200 per month.
Choose the right spending commitment
Regardless of actual resource utilization, AWS Savings Plan users commit to an hourly dollar spend. If an organization commits to $5/hour using Savings Plans, the business owners pay this amount for each hour in the commitment period. If, for some reason, the compute utilization is $3 in a specific hour, the account still pays $5, and any unused commitment amount is not applied to other hours. Any compute usage exceeding the hourly Savings Plans commitment is billed at On-Demand rates.
Based on the hourly-commitment nature of Savings Plans, CTOs must calculate the Savings Plans dollar amount after all compute infrastructure optimizations have been applied in the AWS account or organization. Otherwise, it is possible to end up in a situation where the organization does not use AWS Savings Plan to their full potential.
For example, an organization might purchase a $10/hour commitment based on the costs related to non-optimal compute configurations. After applying optimizations, that same organization might only need a $5/hour commitment. This is a common situation that, unfortunately, can nullify any potential savings when optimizations are not applied before calculating the hourly commitment.
The AWS Compute Optimizer tool identifies and applies compute optimizations before purchasing Savings Plans. For deployments that require more flexibility from the compute infrastructure, Compute Savings Plans offer broader coverage of compute resources than EC2 Savings Plans.
It is also important to calculate the right number of Savings Plans using the baseline utilization throughout the month. It is possible to configure an over-commitment. For example, if an application has spikes that bring utilization to $20/hour in some periods, but the baseline utilization is $5/hour, those spikes can raise the average hourly utilization and cause confusion when purchasing Savings Plans.
There are several tools in the Billing and Cost Management console that can help determine the correct hourly commitment, such as Savings Plans Purchase Analyzer and Savings Plans recommendations.
Ernesto Marquez is owner and project director at Concurrency Labs, where he helps startups launch and grow their applications on AWS. He enjoys building serverless architectures, building data analytics solutions, implementing automation and helping customers cut their AWS costs.
