AWS Cost Optimization for PetraRide & Associated Accounts

Overview

PetraRide operates multiple AWS environments (Production, Development, and related accounts) across several regions. A comprehensive cost optimization assessment was conducted to identify high-impact savings opportunities without affecting system reliability or performance. The focus areas included compute, databases, networking, logging, and long- term commitment models (Reservations & Savings Plans).

Key Challenges Identified

• High compute and database costs driven by non-optimized instance types and runtimes
• Public networking exposure increasing VPC and IP-related charges
• High CloudWatch log ingestion, mainly from Lambda
• Resources running 24/7 in non-production environments
• Idle and underutilized resources (EC2, ALBs, EIPs, ECR images)
• Lack of commitment-based savings (Reservations & Savings Plans)
• Some workloads reaching Extended Support or using non-cost-efficient configurations

Recommended Optimization Actions

1.Compute & Container Optimization (ECS, Lambda, EC2)

• Migrate ECS Fargate services and Lambda functions to ARM (Graviton) where applicable
• Apply rightsizing and auto-scaling
• Clean up stopped EC2 instances and unused EBS volumes
• Introduce scheduling (8 hours/day) for development workloads

Estimated Impact:
High monthly savings with low-to-medium implementation effort
Requires testing in development before production rollout

2.Database Optimization (RDS & Aurora)

• Migrate selected Aurora clusters from Standard to I/O-Optimized
• Apply Graviton-based instance types
• Schedule non-production databases
• Remove old snapshots and upgrade outdated engines
• Evaluate Extended Support upgrades with the customer

Estimated Impact:
One of the highest contributors to annual savings
Low operational risk with minimal downtime

3.Networking Optimization (VPC & Load Balancing)

• Reduce public IP usage (ECS, RDS, EC2)
• Route outbound traffic via NAT Gateway
• Disable “Publicly Accessible” for RDS
• Remove idle Elastic IPs and unused interface endpoints
• Reduce number of ALBs and optimize subnet usage

Estimated Impact:
Moderate savings with improved security posture
Requires coordination but minimal service disruption

4.Logging & Observability Optimization

• Move Lambda logs from CloudWatch to New Relic
• Create CI/CD pipelines for Lambda functions
• Use aliases and versions to ensure safe rollback

Estimated Impact:
Significant reduction in CloudWatch ingestion costs
Joint effort required with customer teams

5.Service Selection Improvements

• Migrate high-traffic REST APIs from API Gateway to ALB
• Replace idle or redundant ALBs with CloudFront where applicable

Estimated Impact:
High cost reduction for request-heavy workloads
Requires testing and domain migration planning

6.Search & Analytics Optimization (OpenSearch)

• Rightsize data and master nodes
• Balance data nodes across AZs
• Apply scheduled runtime in development environments
• Purchase OpenSearch Reserved Instances for stable workloads

7.Commitment-Based Savings

• Purchase:
  • Compute Savings Plans (3 years, no upfront)
  • RDS Reserved Instances
  • OpenSearch Reserved Instances
• Target stable, predictable workloads in production

Estimated Impact:
Very high annual savings
No operational changes required

Estimated Savings Summary

• Production accounts (without reservations): ~$156K/year
• All accounts (without reservations): ~$140K/year
• All accounts (with Savings Plans & Reservations): ~$190K/year

General Best Practices Implemented & Recommended

• AWS Budgets & Budget Alerts
• Cost Dashboards
• Cost Anomaly Detection
• Cost Optimization Hub (already enabled)

Conclusion

This use case demonstrates how a structured, multi-pillar AWS cost optimization strategy — combining rightsizing, ARM adoption, scheduling, service selection, and long-term commitments — can deliver six-figure annual savings while maintaining system stability and scalability.