Your browser doesn't support the features required by impress.js, so you are presented with a simplified version of this presentation.

For the best experience please use the latest Chrome, Safari or Firefox browser.

Profilers

Julian Edwards, Henrik Ingo,
Bruce Lucas, Eric Milkie

MongoDB Performance Tech Talks 2019

# Agenda - What is performance, anyway? - Profilers and other tools - Demos - Julian Edwards: perf & FlameGraphs - Henrik Ingo: gdb & calltree.py - Bruce Lucas: eBPF & t2 - Eric Milkie: Intel VTune
# Correctness * Binary * Given input => output * Must not crash * Any deviation is FAIL * Failures can be reproduced and debugged in 1-3 threads
# Performance * Better / worse. NOT binary. * Holistic: What is total output of all threads? * Profiling * Where does a thread spend time over its lifecycle? * Where does the system spend time, aggregated over all threads?

Optimization question

Last month I spent $450 on Starbucks & $50 on champagne.

What can I do to save money?

 

 

This is profiling!                    

# All the tools * Evergreen * Build * Deploy * Test (or Prod) * Metrics * BF (or SFDC) ticket * **Profiling** * SERVER ticket * git commit
# Demo - Julian Edwards: perf & FlameGraphs - Henrik Ingo: gdb & calltree.py - Bruce Lucas: eBPF & t2 - Eric Milkie: Intel VTune