Overview
AWS reveals that S3, one of the world’s largest cloud storage services, has been quietly rewriting almost everything in its request path using Rust over the years. This migration from the original 2006 codebase represents a major engineering shift driven by performance demands.
Key Takeaways
- Performance-critical systems benefit from gradual rewrites - AWS didn’t rebuild S3 from scratch but incrementally replaced components in the request path with Rust implementations
- Memory safety and performance can coexist - Rust allows AWS to achieve the lowest latency and highest performance while maintaining safety guarantees that garbage-collected languages can’t provide
- Large-scale system migrations happen quietly - Major infrastructure companies often make significant technology changes behind the scenes without announcing them, focusing on results over publicity
- Hardware evolution drives software choices - AWS’s work with cutting-edge hardware requires languages that can extract maximum performance without runtime overhead
Topics Covered
- 0:00 - Hardware Strategy and Trade-offs: Discussion of AWS’s heterogeneous hardware environment and approach to working with latest manufacturer hardware
- 0:15 - S3’s Rust Migration Revealed: AWS engineer reveals that almost everything in S3’s request path has been rewritten in Rust over the years
- 0:30 - Performance-Driven Engineering: Explanation of why AWS chose Rust for S3 - seeking highest performance and lowest latency in critical code paths