AWS re:Invent 2025 - Optimize AWS Costs: Developer Tools and Techniques (DEV318)

AWS re:Invent 2025 - Optimize AWS Costs: Developer Tools and Techniques (DEV318)

Introduction

• Scenario: Unexpected spike in AWS budget leading to panic and manual optimization efforts
• Presenters: Steph Gu (AWS Senior SA) and Kenneth (Enterprise Architect at Betsson, AWS Hero)
• Goals:
  1. Explore cost optimization opportunities and techniques
  2. Leverage AI tools to speed up the optimization process
  3. Implement preventative measures to avoid future waste

Simple Optimizations

1. CloudTrail: Consolidated organization-level CloudTrail to reduce costs from multiple trails.
2. Volumes: Deleted unattached EBS volumes, taking backups where needed for compliance.
3. Networking: Removed unused Elastic Load Balancers, NAT Gateways, and Transit Gateways.
4. Right-Sizing: Used tools like AWS Cost Optimization Hub and AWS Compute Optimizer to identify right-sizing opportunities.
5. Storage: Removed unused S3 data, moved some to Glacier, and implemented S3 Intelligent-Tiering.
6. Graviton: Migrated RDS databases and other services to Graviton-based instances for better performance and cost-efficiency.

Advanced Optimizations

1. CloudWatch:
   • Removed unnecessary log groups and set proper retention policies.
   • Leveraged Infrequent Access storage class to reduce log storage costs.
2. AWS Config:
   • Optimized EKS configuration from continuous to daily for reduced Config logging.
3. Data Transfer:
   • Implemented S3 Gateway Endpoints to avoid data transfer charges between VPCs and S3.
   • Optimized Kubernetes networking to minimize cross-AZ data transfer.
4. Lambda:
   • Migrated Lambda functions to Graviton-based instances.
   • Optimized code to reduce compute requirements and execution time.

Deep Dive Optimizations

1. Network Gateway Optimization:
   • Replaced multiple VPC-level NAT Gateways with a centralized Egress VPC and Transit Gateway.
   • Reduced data transfer costs by 35-45% through S3 Gateway Endpoints and network ACLs.
   • Break-even point for this approach is around 10 VPCs, with up to 8% savings for larger setups.
2. Region Selection for Data Transfer:
   • Chose US East 1 region for a Database Migration Service (DMS) instance instead of São Paulo, saving $2,438 per month.
   • Region pricing can vary significantly, especially for data transfer-heavy services.
3. AWS WAF Optimization:
   • Utilized standard WAF features (e.g., geo-blocking, default rules) to minimize traffic hitting the more expensive premium features.
   • Leveraged the "Challenge" feature over "CAPTCHA" to reduce costs from $4 per 10K requests to $0.40 per million.
   • Optimized WAF capacity by combining CloudFront and origin-level WAF deployments.

Preventing Future Waste

1. Service Control Policies (SCPs):
   • Guardrails set at the organization or OU level to prevent unwanted resource deployments.
   • Examples: Deny use of specific instance types, enforce Graviton usage, restrict regions, mandate tagging.
   • Can be tested and validated using AI tools like Amazon CodeGuru and Kira CLI before deployment.
2. Contextual AI Integration:
   • Leveraged Amazon CodeGuru and Kira CLI to build infrastructure with pre-configured optimization rules.
   • Stored personal preferences and best practices in "context" to guide AI-generated code and recommendations.

Key Takeaways

• Utilize a combination of simple and advanced optimization techniques to reduce AWS costs.
• Leverage AI-powered tools like AWS Cost Optimization Hub, Amazon CodeGuru, and Kira CLI to speed up the optimization process.
• Implement preventative measures like Service Control Policies and contextual AI integration to avoid future waste.
• Focus on optimizing networking, storage, compute, and data transfer to achieve significant cost savings.
• Choose the right AWS regions and services to minimize data transfer and other variable costs.
• Continuously monitor and optimize your AWS environment to maintain cost-efficiency.

Business Impact

• Avoided unexpected budget spikes and the associated panic and manual effort.
• Reinvested cost savings into new features, staff, and other business initiatives.
• Improved overall cost-efficiency and financial predictability of the AWS environment.
• Empowered developers to optimize costs as part of their regular development workflow.
• Established a culture of cost-consciousness and proactive optimization within the organization.

Technical Details and Examples

• Specific cost savings metrics:
  • 35-45% reduction in data transfer costs through S3 Gateway Endpoints and network ACLs.
  • $2,438 monthly savings by deploying a DMS instance in US East 1 instead of São Paulo.
• Leveraged tools like AWS Cost Optimization Hub, Amazon CodeGuru, and Kira CLI for optimization recommendations and automated implementation.
• Implemented a centralized Egress VPC with Transit Gateway to optimize network gateway costs.
• Optimized AWS WAF usage by focusing on standard features and the "Challenge" feature over "CAPTCHA".
• Used Service Control Policies to enforce best practices like Graviton usage, region restrictions, and tagging requirements.