Real Stories of Cost Optimization on AWS
In this post, we will unveil best practices and real-life customer sagas where businesses harnessed the power of AWS to supercharge their savings
Venkatakrishna Pullela
Amazon Employee
Published Sep 1, 2023
Last Modified Dec 18, 2024
I am a Senior AWS Technical Account Manager, and in my role I provide consultative architectural and operational guidance delivered in the context of customers' applications and use-cases to help them achieve the greatest value from AWS. My role is not just about providing technical support but also helping our valued customers optimize their cloud resources to achieve cost-efficiency without sacrificing performance. In this blog post, we'll explore various strategies, tactics, and real-world examples of how I've assisted our large enterprise customers in their cost optimization endeavors. I'll share real-life use cases, metrics, and strategies that have proven effective. Stay tuned for a Bonus strategy at the end.
In the ever-evolving landscape of cloud computing, Enterprises are continually seeking ways to maximize the benefits of AWS while keeping costs in check. Before diving into specific customer cases, let's outline the fundamental framework for AWS cost optimization. It involves several key pillars:
Implementing Cloud Financial Management starts with the customer identifying key stakeholders from Finance and Technology, thus forming a cost optimization function that's responsible to establish and maintain cost awareness internally. A suite of AWS Services can help the customer achieve this, such as AWS Budgets and AWS Cost Anomaly Detection to accurately forecast AWS spend. Tools like AWS Cost Explorer and AWS Billing Console can be leveraged to derive real-time insights into AWS spend , while AWS Cost and Usage reports serve an excellent source to obtain fine-grain usage metrics. Additionally, customers can subscribe to AWS Blogposts to stay relevant to Roadmap items of interest. The key pillars behind implementing Cloud Financial Management include:
- Establish a FinOps Practice: Create a framework for managing cloud costs that includes budget planning, forecasting, and cost allocation. Empower teams to make informed trade-offs between cost, speed, and quality.
- Use AWS Budgets and Cost Anomaly Detection: Set up budgets to monitor and manage costs in real time, and detect anomalous spending patterns before they escalate.
- Implement Tagging Strategies: Adopt a comprehensive resource tagging strategy for cost attribution and accountability. This helps identify cost drivers by team, project, or environment.
- Leverage AWS Cost Management Tools: Utilize tools such as the AWS Cost Explorer, AWS Billing Console, and Trusted Advisor to analyze spending patterns, set alerts, and take corrective actions.
CFM ensures that cloud costs align with business objectives while enabling continuous cost optimization.
Expenditure and Usage awareness among internal teams, can be enabled via a unified Account structure with AWS Organizations, allowing customers to seamlessly allocate costs and usage to the various Business Units within the company. Customers can setup and enable Cost Allocation tags as part of their charge-back strategy to hold internal product teams accountable for their AWS Costs and Usage. Broadly speaking, Expenditure and Usage Awareness can be involve:
- Enable AWS Cost Explorer and Reports: Use the AWS Cost Explorer to visualize cost and usage trends over time. Dive into granular details, such as service-level spending or account-level charges.
- Set Up Detailed Billing and Resource Grouping: Enable detailed billing with cost and usage reports for a comprehensive breakdown of your expenses. Use resource grouping to aggregate costs by function or team.
- Monitor Reserved Instance and Savings Plan Utilization: Ensure your reserved resources are being utilized optimally. Regularly review commitment-based plans to align with usage patterns.
- Educate Teams on Cost Awareness: Foster a culture where teams actively monitor their resource usage and take steps to eliminate waste.
Expenditure and usage awareness equips your organization to spot inefficiencies and adjust usage to achieve cost-effective outcomes.
Using purpose-built resource types for specific workloads is going to be key to cost savings. Other opportunities includes, Instance resizing for both Cost and Performance or leveraging managed services such as Amazon Aurora, Amazon ElastiCache and Amazon DynamoDB to minimize the overhead of managing and maintaining the servers. Customers can also take advantage of AWS' cost-effective pricing options such as Savings Plans, Reserved Instances or Spot for their workloads, saving significant amount of costs as compared to OnDemand. Some key strategies include:
- Leverage Spot Instances: Take advantage of AWS Spot Instances for fault-tolerant and flexible workloads, such as batch processing and big data analysis, to achieve significant savings.
- Right-Sizing Resources: Regularly assess your EC2 instance types, storage volumes, and database configurations to ensure they match workload requirements. Avoid over-provisioning.
- Use Managed Services: Shift to AWS-managed services like Amazon RDS, AWS Lambda, or Amazon ECS to minimize operational overhead and reduce costs.
- Employ Storage Tiering: Use storage classes such as Amazon S3 Intelligent-Tiering or Glacier for data that doesn’t require frequent access.
By aligning your resource choices with workload demands, you can achieve a cost-effective and scalable cloud infrastructure.
It is paramount for customers to keep their operating costs low. To facilitate that, customers can consider adopting auto-scaling across key AWS services such as Amazon EC2, Amazon DynamoDB and Amazon ElastiCache. This ensures that the customer only pays for the computing resources they need, and not have to overpay. Consider the following strategies:
- Implement Auto-Scaling: Use AWS Auto Scaling to dynamically adjust compute and storage resources based on real-time demand. This prevents under-utilization and reduces over-provisioning costs.
- Leverage Queue-Based Architectures: For workloads with variable traffic, queue-based systems like Amazon SQS can help smooth out demand spikes and ensure efficient resource utilization.
- Use Predictive Scaling: Take advantage of AWS predictive scaling to anticipate future demand based on historical data and adjust resources proactively.
- Schedule Resource Usage: Identify non-critical workloads and schedule their usage during off-peak hours using AWS Instance Scheduler or EventBridge rules.
Effective demand and supply management ensures that resources are optimally aligned with workload requirements, improving both cost-efficiency and system performance.
In AWS, you optimize over time by reviewing new services and implementing them in your workload. As AWS releases new services and features, it is a best practice to review your existing architectural decisions to ensure that they remain cost effective. As your requirements change, be aggressive in decommissioning resources, components, and workloads that you no longer require. Key principles include:
- Regularly Review and Refactor: Conduct periodic reviews of your architecture using AWS Well-Architected Tool to identify cost optimization opportunities. Refactor workloads to take advantage of new AWS services and pricing models.
- Adopt Elastic and Serverless Architectures: Serverless solutions like AWS Lambda and managed services like Fargate allow you to pay only for what you use, reducing costs over time.
- Monitor and Adapt to New AWS Features: Stay updated with AWS announcements to leverage new features or services that could offer cost benefits.
- Optimize Data Transfer Costs: Minimize cross-region data transfers by architecting for locality and using Amazon CloudFront for caching and delivery.
Continuous optimization ensures your cloud spending remains efficient and aligned with evolving business goals.
A Digital AdExchange realized over 60% in cost savings by adopting AWS EC2 Spot for their real-time bidding application, while maintaining the same throughput. The customer updated their application to use disparate Instance Types (~20) and over 50% of those Instance types were used in the spot fleets that replaced the original auto-scaling groups, thus providing a significant amount of resistance to fluctuations in the spot market. The customer also implemented a log recovery pipeline to combat the problem of losing log data on terminated spot instances. The customer had unpredictable traffic patterns but needed cost predictability. So, in addition to Spot, we helped them implement a mixed Reservations strategy, combining EC2 Savings Plans for steady-state workloads and Compute Savings Plans for flexibility. While the EC2 Instance Savings Plans provided the customer with over 65% price benefits, the Compute Savings Plans unlocked over 60% in potential cost savings. This led to annual cost savings of over $1 million. Furthermore, Advertising technology (ad tech) companies use Amazon DynamoDB to store various kinds of marketing data, such as user profiles, user events, clicks, and visited links. Some of the use cases include real-time bidding (RTB), ad targeting, and attribution. These use cases require a high request rate (millions of requests per second), low and predictable latency, and reliability. As a fully managed service, DynamoDB allows ad tech companies to meet all of these requirements without having to invest resources in database operations. With accelerated DynamoDB costs Month-over-month, the customer quickly adopted DynamoDB Reserved Capacity to receive a significant discount over Provisioned Capacity (Read and Write) and this helped the customer save an estimated $1M annually. Lastly, the customer was also able to save ~$300k with ElastiCache RIs and $150k with OpenSearch RIs.
One of our largest social media customers experienced a sudden but sustained spike in their AWS costs for a specific Business Unit. We implemented a comprehensive tagging strategy, enabling them to allocate costs with precision. This led to better cost accountability and improved budget management.
We also set up AWS Cost and Usage reports to provide granular insights into Cost and Usage, and we were able to identify the key driver behind the cost increase. This proactive approach allowed the company to identify and address cost anomalies in real-time, saving them thousands of dollars. Additionally, it helped the customer gain visibility into escalating Compute and Database costs. AWS Trusted Advisor is a service that inspects your AWS environment and provides recommendations for optimizing your resources. For one of our largest social media customers, we utilized AWS Trusted Advisor, which identified Underutilized and Idle resources that could potentially be terminated savings the customer $200k monthly across Amazon EC2, EBS, RDS and Load Balancers. Furthermore, the tool also provided recommendations around. Lastly with AWS Compute Optimizer, the customer was able to uncover $60k in monthly savings achieved through EC2 Rightsizing recommendations.
We also set up AWS Cost and Usage reports to provide granular insights into Cost and Usage, and we were able to identify the key driver behind the cost increase. This proactive approach allowed the company to identify and address cost anomalies in real-time, saving them thousands of dollars. Additionally, it helped the customer gain visibility into escalating Compute and Database costs. AWS Trusted Advisor is a service that inspects your AWS environment and provides recommendations for optimizing your resources. For one of our largest social media customers, we utilized AWS Trusted Advisor, which identified Underutilized and Idle resources that could potentially be terminated savings the customer $200k monthly across Amazon EC2, EBS, RDS and Load Balancers. Furthermore, the tool also provided recommendations around. Lastly with AWS Compute Optimizer, the customer was able to uncover $60k in monthly savings achieved through EC2 Rightsizing recommendations.
AWS Graviton is a game-changer for cost optimization and performance in Cloud Computing. Graviton instances, powered by Arm-based processors, offer a compelling value proposition. Graviton instances typically come at a lower price point compared to their x86 counterparts, making them a cost-effective choice. They also consume less power and have a smaller carbon footprint, reducing operational costs and contributing to an organization's sustainability goals. For businesses with demanding workloads, this can translate into significant savings over time. Along with the cost benefits, the Graviton instances are also designed for performance and efficiency making it a lucrative choice for many types of workloads, especially those that are highly parallelizable, like web servers and microservices. A major streaming company whose key workloads are hosted on AWS were able to assess $1M in monthly net savings (20%) by adopting Graviton across Amazon EC2, RDS, ElastiCache and OpenSearch. This allowed the customer actively re-invest the realized savings into other growth areas such as Generative AI that helped accelerate their business outcomes.
As a AWS Technical Account Manager, I'm committed to staying at the forefront of these developments, ensuring that our customers have access to the latest cost optimization strategies and technologies. Cost optimization on AWS isn't a one-time task but a continuous journey. By leveraging rightsizing, reserved instances, spot instances, lifecycle policies, and effective cost allocation, we've helped numerous enterprise customers unlock substantial savings while maintaining or even improving their cloud workloads. The key takeaway is that cost optimization on AWS is achievable with the right strategies, tools, and a commitment to ongoing monitoring and improvement. As your partner in the cloud, I'm here to guide you on this journey, ensuring that your AWS infrastructure remains cost-efficient, agile, and aligned with your business objectives. In the world of AWS, the path to cost optimization is paved with data-driven insights, smart resource management, and a collaborative partnership between AWS and our valued customers. Together, we'll continue to optimize and thrive in the cloud.
Any opinions in this post are those of the individual author and may not reflect the opinions of AWS.