Optimizing Cost and Performance in Amazon RDS and Amazon Aurora

Journey with Amazon RDS

Challenge 1: Increased Workload

• Solution: Used instance right-sizing and leveraged the latest Graviton innovation to achieve improved price-performance.

Challenge 2: Noisy Neighbors

• Solution: Offloaded reporting queries to read replicas to eliminate interference with the core customer application.

Challenge 3: I/O Latency

• Solution: Performed storage right-sizing by moving from GP2 to IO2 Block Express to achieve sub-millisecond I/O latency.

Challenge 4: Slow Reporting Queries

• Solution: Used RDS Optimized Reads feature with the latest Graviton instances to improve query performance by 2x without significant cost increase.

Challenge 5: High Backup Costs

• Solution: Reduced backup costs by 30% by using a cost-efficient long-term backup approach, including snapshot export to S3 and logical backups.

Importance of Visibility and Detection

• Discussed the importance of using tools like CloudWatch, Performance Insights, and Enhanced Monitoring to identify and address performance and cost challenges.
• Announced the new CloudWatch Database Insights feature, which provides a unified view of database monitoring and the ability to correlate database performance with application performance.

Journey with Amazon Aurora

Challenge 1: Burst I/O Costs

• Solution: Leveraged Aurora I/O Optimized pricing option to improve cost predictability and reduce I/O costs by up to 28%.

Challenge 2: Noisy Neighbors

• Solution: Introduced read replicas to offload reporting queries and improve performance without impacting the core application.

Challenge 3: Testing Environment Costs

• Solution: Used Aurora Fast Clones to create test environments at low cost, achieving a 90% storage cost reduction and 38% compute cost reduction.

Challenge 4: Global Resilience and Low Latency

• Solution: Implemented Aurora Global Database to enable low-latency reads in multiple regions and provide global disaster recovery.

Challenge 5: Scalability Limits

• Solution: Leveraged Amazon Aurora Serverless v2 (Postgres Limitless) to automatically scale storage and throughput as needed, without capacity constraints.

Key Takeaways

• Optimizing cost and performance are complementary tasks that should be addressed together.
• Both Amazon RDS and Amazon Aurora provide common tools, such as Performance Insights and Enhanced Monitoring, to help identify and address performance and cost challenges.
• Visibility and detection are crucial, and the new CloudWatch Database Insights feature can provide a unified view and correlation between database and application performance.
• Leveraging the latest innovations in Amazon RDS and Amazon Aurora, such as Graviton instances, I/O Optimized pricing, and Aurora Serverless v2, can help achieve significant cost savings and performance improvements.