slides: static, benchmarking, tuning
sar, perf-tools, bcc/BPF:
Images license: creative commons Attribution-ShareAlike 4.0.
This page links to various Linux performance material I've created, including the tools maps on the right. The first is a hi-res version combining observability, static performance tuning, and perf-tools/bcc (see discussion). The remainder were designed for use in slide decks and have larger fonts and arrows, and show: Linux observability tools, Linux benchmarking tools, Linux tuning tools, and Linux sar. For even more diagrams, see my slide decks below.
- perf_events: perf one-liners, examples, visualizations.
- eBPF tools: eBPF tracing tools and examples with bcc.
- perf-tools: perf analysis tools using ftrace (github).
- bcc: perf analysis tools using eBPF (github).
- ktap: one-liners, examples, and scripts.
- Flame Graphs: using perf, SystemTap, and ktap.
- Linux Performance Analysis in 60,000 Milliseconds shows the first ten commands to use in an investigation (video, PDF). Written by myself and the performance engineering team at Netflix (2015).
- My post Performance Tuning Linux Instances on EC2 includes the tunables we're using at Netflix (2015).
- A post on Linux Load Averages: Solving the Mystery, explaining what they are and why they include the uninterruptible sleep state (2017).
- A gdb Debugging Full Example (Tutorial), including the use of some perf/debugging tools (2016).
- Posts about eBPF and bcc (2015-7):
Linux eBPF (2015)
bcc: Taming Linux 4.3+ Tracing Superpowers
tcpconnect and tcpaccept for Linux (bcc)
Linux eBPF Stack Trace Hack (bcc) (2016)
Linux eBPF Off-CPU Flame Graph (bcc)
Linux Wakeup and Off-Wake Profiling (bcc)
Linux chain graph prototype (bcc)
Linux eBPF/bcc uprobes
Linux BPF/bcc Road Ahead
Ubuntu Xenial bcc/BPF
Linux bcc/BPF Tracing Security Capabilities
Linux MySQL Slow Query Tracing with bcc/BPF
Linux bcc/BPF ext4 Latency Tracing
Linux bcc/BPF Run Queue (Scheduler) Latency
Linux bcc/BPF Node.js USDT Tracing
Linux bcc tcptop
Linux 4.9's Efficient BPF-based Profiler
DTrace for Linux 2016
Linux 4.x Tracing Tools: Using BPF Superpowers
Linux bcc/BPF tcplife: TCP Lifespans
Golang bcc/BPF Function Tracing (2017)
7 BPF tools for performance analysis on Fedora
- My lwn.net article Ftrace: The Hidden Light Switch shows a use case for Linux ftrace (Aug, 2014).
- Posts about ftrace-based perf-tools (2014-5):
- Posts about perf-based perf-tools: perf Hacktogram.
- Posts about perf_events (2014-7):
- In The PMCs of EC2: Measuring IPC I showed the new Performance Monitoring Counter (PMC) support in the AWS EC2 cloud (2017).
- CPU Utilization is Wrong: a post explaining the growing problem of memory stall cycles dominating the %CPU metric (2017).
- A post about Linux 4.7 Hist Triggers (2016).
- The blog post strace Wow Much Syscall discusses strace(1) for production use, and compares it to advanced tracing tools (2014).
- USE Method: Linux Performance Checklist; also see the USE Method page for the description of this methodology.
- Systems Performance: Enterprise and the Cloud (Prentice Hall, 2013) uses Linux distributions as the primary example.
In rough order of recommended viewing or difficulty, intro to more advanced:
1. Linux Systems Performance (PerconaLive 2016)
This is my summary of Linux systems performance in 50 minutes, covering six facets: observability, methodologies, benchmarking, profiling, tracing, and tuning. It's intended for people who have limited appetite for this topic.
For a lot more information on observability tools, profiling, and tracing, see the talks that follow.
2. Linux Performance Tools (Velocity 2015)
At Velocity 2015, I gave a 90 minute tutorial on Linux performance tools, summarizing performance observability, benchmarking, tuning, static performance tuning, and tracing tools. I also covered performance methodology, and included some live demos. This should be useful for everyone working on Linux systems. If you just saw my PerconaLive2016 talk, then some content should be familiar, but with many extras: I focus a lot more on the tools in this talk.
This was similar to my SCaLE11x and LinuxCon talks, however, with 90 minutes I was able to cover more tools and methodologies, making it the most complete tour of the topic I've done. I also posted about it on the Netflix Tech Blog.
3. How Netflix Tunes EC2 Instances for Performance (AWS re:Invent, 2017)
Instead of performance observability, this talk is about tuning. I begin by providing Netflix background, covering instance types and features in the AWS EC2 cloud, and then talk about Linux kernel tunables and observability.
4. Container Performance Analysis (DockerCon, 2017)
At DockerCon 2017 in Austin, I gave a talk on Linux container performance analysis, showing how to find bottlenecks in the host vs the container, how to profiler container apps, and dig deeper into the kernel.
5. Broken Linux Performance Tools (SCaLE14x, 2016)
At the Southern California Linux Expo (SCaLE 14x), I gave a talk on Broken Linux Performance Tools. This was a follow-on to my earlier Linux Performance Tools talk originally at SCaLE11x (and more recently at Velocity as a tutorial). This broken tools talk was a tour of common problems with Linux system tools, metrics, statistics, visualizations, measurement overhead, and benchmarks. It also includes advice on how to cope (the green "What You Can Do" slides).
6. Using Linux perf at Netflix (Kernel Recipes, 2017)
At Kernel Recipes 2017 I gave an updated talk on Linux perf at Netflix, focusing on getting CPU profiling and flame graphs to work. This talk includes a crash course on perf_events, plus gotchas such as fixing stack traces and symbols when profiling Java, Node.js, VMs, and containers.
There's also an older version of this talk from 2015, which I've posted about.
7. Linux Performance Analysis: New Tools and Old Secrets (ftrace) (LISA 2014)
At USENIX LISA 2014, I gave a talk on the new ftrace and perf_events tools I've been developing: the perf-tools collection on github, which mostly uses ftrace: a tracer that has been built into the Linux kernel for many years, but few have discovered (practically a secret).
The perf-tools collection, inspired by my earlier DTraceToolkit, provides advanced system performance analysis tools for Linux. Each tool has a man page and example file. They are unstable and unsupported, and they currently use shell scripting, hacks, and the ftrace and perf_events kernel tracing frameworks. They should be developed (and improved) as the Linux kernel acquires more tracing capabilities (eg, eBPF).
In a post about this talk, I included some more screenshots of these tools in action.
8. Give me 15 minutes and I'll change your view of Linux tracing (LISA, 2016)
I gave this demo at USENIX/LISA 2016, showing ftrace, perf, and bcc/BPF. A video is on youtube.
9. Performance analysis superpowers with Linux eBPF (O'Reilly Velocity, 2017)
This talk covers using enhanced BPF (aka eBPF) features added to the Linux 4.x series for performance analysis, observability, and debugging. The front-end used in this talk is bcc (BPF compiler collection), an open source project that provides BPF interfaces and a collection of tools.
10. Performance Checklists for SREs (SREcon, 2016)
At SREcon 2016 Santa Clara, I gave the closing talk on performance checklists for SREs (Site Reliability Engineers). The later half of this talk included Linux checklists for incident performance response. These may be useful whether you're analyzing Linux performance in a hurry or not.
11. What Linux can learn from Solaris performance and vice-versa (SCaLE12x, 2014)
At SCaLE 12x, I gave the keynote on What Linux can learn from Solaris performance and vice-versa. This drew on my prior experiences doing head to head comparisons, and work for the Systems Performance book. I'd never seen a good talk comparing performance features of both, I suspect in part because it's hard to know them both in enough depth, and also hard to choose from the many differences which should be highlighted.
Update 2017: This talk is now a little out of date, as Linux has come a long way in the last three years (Linux 3.13 to Linux 4.15), whereas illumos/Solaris have not. Solaris is also believed to have died this year, so I wrote the post Solaris to Linux Migration 2017.
Other resources (not by me) I'd recommend for the topic of Linux performance:
- Performance analysis & tuning of Red Hat Enterprise Linux - 2015 Red Hat Summit (video 2hrs): this is a great and in-depth tour of Linux performance tuning that should be largely applicable to all Linux distros.
- Linux Instrumentation: slides from a great talk in June 2010 by Ian Munsie, which summarizes the different Linux tracers very well. If you're trying to understand all the tracers and frameworks, this is worth studying (keeping in mind it's from 2010).
- Julia Evans blog has many posts about many topics, including performance tools.