Many organizations today are increasingly reliant on cloud infrastructure, particularly on platforms like Amazon Web Services (AWS), which often consumes a substantial portion of their IT budgets. In light of this, AWS Savings Plans emerge as a potential solution for minimizing costs effectively.
Optimizing cloud costs within AWS involves several best practices, including meticulous cost budgeting, continuous monitoring, and infrastructure optimizations. AWS has developed specialized tools to help organizations implement FinOps best practices. By utilizing AWS Savings Plans, businesses can unlock substantial discounts on compute utilization expenses, thereby enhancing the overall Return on Investment (ROI) from their cloud deployments.
AWS Savings Plans facilitate cost reductions through a commitment to a fixed hourly spend on certain usage types associated with AWS’s compute infrastructure. Organizations can choose from one or three-year commitment options, with flexibility in payment structures available as monthly, partial-upfront, or all-upfront payments. Generally, the longer the commitment period and the higher the upfront payment, the more significant the cost reductions applicable to various resources throughout an AWS Organization.
Savings Plan Options to Consider
AWS offers a variety of Savings Plans to cater to different organizational needs:
- Compute Savings Plans: These encompass multiple compute services, including AWS Lambda, EC2 instances, and Fargate, and are valid across different AWS regions.
- EC2 Instance Savings Plans: Commitments made under this plan are specific to particular instance families and regions within AWS.
- SageMaker Savings Plans: These cover all instance families and regions supported by AWS SageMaker AI.
- Database Savings Plans: This option applies to all regions, instance families, and database engines supported by Amazon RDS.
A notable distinction exists between Compute Savings Plans and EC2 Instance Savings Plans in regard to flexibility. While the latter is restricted to specific instance families, Compute Savings Plans afford the user flexibility by covering various compute resources, including Lambda functions and containers deployed via Fargate. However, organizations should consider that the cost reductions from Compute Savings Plans may be marginally lower, prompting a careful evaluation of their flexibility requirements. For recently launched infrastructure, maintaining this flexibility is crucial for adjusting configurations while still gaining the offered savings.
When contemplating the IT budget for AWS Savings Plans, Chief Technology Officers (CTOs) should be aware of the considerable savings opportunities available. For example, EC2 Savings Plans can yield savings of up to 72%, compared to 66% with Compute Savings Plans. Furthermore, SageMaker AI Savings Plans can offer savings up to 64%, while Database Savings Plans can provide up to 35%. It’s essential to clarify the types of compute utilization that each Savings Plan applies to and to decide whether upfront payment or monthly payment is more suitable for the organization’s financial strategy.
The difference in cost reduction between monthly and all-upfront payments is considerable. For instance, a no-upfront one-year Compute Savings Plan may offer approximately 26% savings based on the usage type, whereas a three-year all-upfront commitment can yield savings of up to 66%. While upfront payments may yield higher savings, organizations must be wary of the burden that high upfront costs can impose. For example, an upfront commitment of $10/hour over three years could require a singular payment nearing $263,000, whereas opting for a no-upfront plan would involve approximately $7,200 monthly payments.
Choosing the Right Spending Commitment
When utilizing AWS Savings Plans, organizations commit to an hourly dollar expenditure, irrespective of their actual resource usage. For instance, if an organization commits to a $5/hour plan, they will incur that cost every hour during the commitment period, even if their compute expenditure is only $3 in a given hour. Any unused portion of the committed amount does not roll over to future hours, and any compute usage exceeding this commitment is charged at On-Demand rates.
This hourly commitment structure necessitates that CTOs diligently assess the required dollar commitment after all optimizations have been applied to their AWS infrastructure. Failure to undertake this assessment may lead to a suboptimal usage of AWS Savings Plans, resulting in lost potential savings. An organization might, for instance, establish a commitment based on non-optimized configurations. After applying optimizations, they may only need a lower hourly rate, negating the benefits of their original commitment.
To aid in this analysis, the AWS Compute Optimizer tool identifies and applies compute optimizations before a purchase of Savings Plans. For deployments that necessitate flexibility, Compute Savings Plans cover a wider range of compute resources compared to EC2 Savings Plans.
Determining the correct quantity of Savings Plans remains pivotal, and it entails evaluating the baseline utilization throughout the month. Businesses risk over-committing if they factor in occasional usage spikes that escalate average utility rates. For example, if an application experiences temporary spikes, raising utilization to $20/hour, but its baseline requirement is only $5/hour, confusion may ensue during the purchasing of Savings Plans.
Within the Billing and Cost Management console, there are multiple tools available to aid in defining the appropriate hourly commitment. These include the Savings Plans Purchase Analyzer and Savings Plans recommendations, essential for making informed decisions regarding cloud cost management.
Ernesto Marquez is the owner and project director at Concurrency Labs, where he assists startups in launching and expanding their applications on AWS. He specializes in building serverless architectures, creating data analytics solutions, implementing automation, and helping clients effectively manage their AWS expenses.