Deterministic profiling is useful for benchmarking low level libraries and language runtimes (e.g. Victor Stinner's recent work on the Python benchmark suite), but statistical profiling has proven to be more practical for analysing performance bottlenecks in managed web services.
Resources like http://www.brendangregg.com/flamegraphs.html are good for understanding how this works in theory, but I'd love to have a talk from someone local that has actually used these tools to help resolve their own performance challenges.
Deterministic profiling is useful for benchmarking low level libraries and language runtimes (e.g. Victor Stinner's recent work on the Python benchmark suite), but statistical profiling has proven to be more practical for analysing performance bottlenecks in managed web services.
Resources like http://www.brendangregg.com/flamegraphs.html are good for understanding how this works in theory, but I'd love to have a talk from someone local that has actually used these tools to help resolve their own performance challenges.