Sonam Gupta and Kunal Nawale, SigLens
Abstract:
Optimizing performance isn’t about guessing — it’s about measuring. Yet, many teams skip systematic profiling and jump straight into rewriting code or scaling infrastructure. The result? Wasted time, minimal gains, and recurring issues. In this talk, we’ll break down five high-impact tips to make profiling a practical, results-driven part of your performance workflow. You’ll learn:
- How to choose the right profiler for your language and environment.
- Where performance bottlenecks really hide (hint: it’s not always your code).
- What to look for in flame graphs, CPU/memory profiles, and I/O traces.
- How to set up lightweight, continuous profiling without slowing down production.
- Ways to prioritize performance fixes that actu…
Sonam Gupta and Kunal Nawale, SigLens
Abstract:
Optimizing performance isn’t about guessing — it’s about measuring. Yet, many teams skip systematic profiling and jump straight into rewriting code or scaling infrastructure. The result? Wasted time, minimal gains, and recurring issues. In this talk, we’ll break down five high-impact tips to make profiling a practical, results-driven part of your performance workflow. You’ll learn:
- How to choose the right profiler for your language and environment.
- Where performance bottlenecks really hide (hint: it’s not always your code).
- What to look for in flame graphs, CPU/memory profiles, and I/O traces.
- How to set up lightweight, continuous profiling without slowing down production.
- Ways to prioritize performance fixes that actually improve user experience. Whether you’re debugging a slow backend or shaving milliseconds off a critical path, these techniques will help you go from vague complaints to concrete wins — backed by data, not guesswork.
