AWS re:Invent 2025 - Boost performance and reduce costs in Amazon Aurora and Amazon RDS (DAT312)

Source: Dev.to

Overview

AWS re:Invent 2025 – Boost performance and reduce costs in Amazon Aurora and Amazon RDS (DAT312)

In this session, Principal Database Solutions Architect Pini Dibask demonstrates performance and cost‑optimization strategies for Amazon RDS and Aurora using a fictional company called AnyCompany. The talk covers three main cost dimensions—compute, storage, and backup—and highlights techniques such as:

• CloudWatch Database Insights for observability
• SQL optimization with proper indexing
• Instance right‑sizing with Graviton processors (up to 46 % cost reduction)
• Leveraging read replicas for workload separation
• Using io2 Block Express for sub‑millisecond latency
• Optimized reads with local NVMe SSD for complex queries

For Aurora, the session explains:

• Aurora I/O‑Optimized (≈ 23 % savings)
• Tiered cache capabilities (≈ 90 % cost reduction)
• Fast clones combined with Aurora Serverless for test environments (≈ 90 % storage savings)
• Aurora Global Database for multi‑region deployments

Key message: cost and performance optimization are complementary; the right tooling and architectural choices can improve both simultaneously.

Introduction – Performance and Cost Optimization Journey with AnyCompany

“Welcome! Today we’ll dive into performance and cost optimization in Amazon Aurora and Amazon RDS through the story of a fictional company, AnyCompany. My goal is that you walk away with actionable insights you can apply to your own database environments.”
— Pini Dibask, Principal Database Solutions Architect, AWS

Amazon RDS & Aurora Primer

• Amazon RDS is a fully managed relational database service that handles upgrades, backups, provisioning, and disaster recovery, letting you focus on application innovation.
• Amazon Aurora is AWS’s cloud‑native relational engine, compatible with MySQL and PostgreSQL, offering enterprise‑grade security, availability, and reliability with the cost‑effectiveness of open‑source databases.
• Aurora is the fastest‑growing AWS service, serving hundreds of thousands of customers.
• The session focuses on Aurora Provisioned (instance‑based) and Aurora Serverless; Aurora DSQL is not covered.

RDS Cost Dimensions

RDS clusters incur costs across several dimensions:

The presentation illustrates each dimension with examples from AnyCompany.

AnyCompany’s First Challenge – CPU Spikes & CloudWatch Database Insights

Background

AnyCompany powers e‑commerce sellers with generative AI, running its application layer on Amazon EKS and its operational database on Amazon RDS. As a fast‑growing startup, they needed to balance performance with cost.

Problem

Sudden CPU spikes to 100 % caused by inefficient SQL statements.

Solution Highlights

1. Enable CloudWatch Database Insights – Provides real‑time visibility into query performance, CPU usage, and wait events.
2. Identify and Optimize Poorly Performing Queries – Use the insights to pinpoint high‑cost queries and add appropriate indexes or rewrite the SQL.
3. Right‑size Compute – After optimization, evaluate whether the current instance class is still required; consider migrating to a Graviton‑based instance for up to 46 % cost savings.

Additional Techniques (Brief Overview)

• Read Replicas – Offload read‑heavy workloads, improve latency, and enable geographic scaling.
• io2 Block Express – Provides sub‑millisecond storage latency for demanding workloads.
• Optimized Reads with Local NVMe SSD – Accelerates complex query processing.
• Aurora I/O‑Optimized – Reduces I/O‑related costs by up to 23 %.
• Tiered Cache – Achieves up to 90 % cost reduction for cached workloads.
• Fast Clones + Aurora Serverless – Enables rapid, cost‑effective test environments with up to 90 % storage savings.
• Aurora Global Database – Supports multi‑region deployments with low‑latency reads.

Takeaways

• Cost and performance are not mutually exclusive; strategic use of AWS tooling and architecture can improve both.
• Observability (CloudWatch Database Insights) is the first step to identifying inefficiencies.
• Right‑sizing and modern architectures (Graviton, Serverless, Global Database) deliver substantial savings without sacrificing performance.