Performance Engineering: Profiling and Making Apps Fast by Default
Summary
Den Odell joins the Real Python Podcast to discuss his book, *Fast by Default*, sharing practical performance engineering strategies applicable across any tech stack. With 25 years of experience, Odell emphasizes integrating performance budgets and profiling into the early development phases rather than treating speed as an afterthought. The conversation covers the trade-offs of using libraries, the importance of measuring perceived performance, and how to communicate the business value of optimization to stakeholders.
Topics discussed
Introduction to Episode 310 and guest Dan Odell
Performance as a full-stack system, not just frontend
Dan's background: from web apps to performance focus
Integrating performance into the development process
The temptation of new tools vs. core browser fundamentals
AI coding agents and the rise of high-level frameworks
Library bloat and the importance of understanding trade-offs
Defining performance budgets based on business value
Managing performance debt and documenting exceptions
Case study: Handling slow API responses in checkout
Origin of the book 'Fast by Default'
Performance budgets in Python and web profiling tools
Using browser developer tools for granular profiling
The importance of early measurement and shift-left
Sponsor segment: Profiling Performance in Python course
Book structure: Concepts, ROI, and business impact
Defining and enforcing performance budgets in CI
Team culture: Performance as a shared responsibility
Real-world testing and perceived performance latency
Monoliths vs. Microservices: Architectural trade-offs
Automating performance checks and PR templates
Handling pushback and managing performance debt
Rethinking load testing: Interaction speed vs. volume
AI code review agents and their limitations
Micro-optimizations and compiler performance insights
Closing thoughts: WebAssembly, sandboxing, and music
Listen ad-free on Castria