Explore Linux systems performance in this comprehensive 41-minute conference talk from LISA19. Dive into six crucial areas: observability tools, methodologies, benchmarking, profiling, tracing, and tuning. Learn practical recipes for performance analysis using tools like vmstat, mpstat, and iostat. Gain insights into complex topics such as profiling with perf_events and tracing with Ftrace, bcc/BPF, and bpftrace/BPF. Discover what's essential for developers, operations professionals, and sysadmins working with Linux in bare metal or cloud environments. Benefit from expert advice on prioritizing performance knowledge and techniques to optimize your Linux systems effectively.
Overview
Syllabus
Introduction
NBStat
PMC Arch
Curve
CP dist
Systems Performance
Load Averages
Top
Htop
VMStat
Free
Perf
TCP Dump
Netstat
SS Slabtop
Page Cache
Containers
Show Boost
Static Performance Tuning
Methodology
Linux Performance Analysis
Profiling
Flame graphs
BPF
Flamescope
Perfect Profile
Tracing
Tracing Stack
Trace
HD for slower
File System
BPF Trace
CPU Analysis
Netflix Tuning
Queue Discs
Summary
Taught by
USENIX
