workshop.txt

Module 1 - Principles
  Who were Benford, Pareto, Zipf and Little?
  Why is premature optimization bad?
  What is benchmarking/profiling?
  How to benchmark an HTTP service
  How do I know how many servers I need to service a given load?
  Little's Law
  Exercise 1: Demonstrate Little's Law by benchmarking a demo application.
Module 2 - Benchmarking
  Benchmark and benchmark/ips
  What can you benchmark?
  What's a good benchmark?
  Pitfalls
  Things to watch out for
  Exercise 2: Benchmark a common feature in your application
Lunch break
Module 3 - Profiling CPU
  Profiling in real time vs production
  Why profiling isn't free
  Profiling workflow (scientific method)
  How profilers lie
  Profiler modes
  CPU time
  Wall Time
  Process time
  Tracers
  Samplers
  How to use ruby-prof
  Exercise 3: Profile a library. options: dalli, raven-ruby, hippie_csv.
Module 4 - Profiling Memory
  Why we profile memory
  How Ruby GC works in 5 minutes
  The GC Module
  Building your own profiler with ObjectSpace
  More ObjectSpace tricks
  What triggers GC?
  Memory myth 1: memory goes down, right?
  Heap fragmentation
  GC tracing
  derailed benchmarks and profiling startup
  oink
  Exercise 4: Let's profile your application!
Module 5 - Profiling the Web Request
  rack-mini-profiler
  Profiling in production-like environments
  Auth with RMP
  Understanding the speed badge
    SQL
    time
    Outliers?
  Flamegraphs
  GC Profiler
  Exception profiling
  Exercise: profile a slow controller action with RMP
Module 6 - Application Metrics
  Why it's fast on your machine
  Development mode settings
  Latency
  Javascript
  System config
  Virtualization
  What's "fast" for Rails
  Applying the 80/20 rule
  Exercise 6: A 10-minute introduction to profiling front-ends with Google Chrome DevTools