Bio

I hate writing bios. From time to time, conference organizers, journalists, and publishers ask me for a bio of a certain length ("50-75 words, "around 100 words", etc), and so I'll cache my most recent ones here. This page also has my upcoming talks.

Very Short

Computer performance analyst, kernel engineer, author, speaker, inventor. Solves hard problems.

Short (<75 words)

Brendan Gregg is a senior performance architect at Netflix, where he does large scale computer performance design, evaluation, analysis, and tuning. He is the author of Systems Performance published by Prentice Hall, and received the USENIX LISA Award for Outstanding Achievement in System Administration. He has previously worked as a performance and kernel engineer, and has created performance analysis tools included in multiple operating systems, as well as visualizations and methodologies.

Medium (around 100 words)

Brendan Gregg is a senior performance architect at Netflix, where he does large scale computer performance design, evaluation, analysis, and tuning. He is the author of multiple technical books including Systems Performance published by Prentice Hall, and received the USENIX LISA Award for Outstanding Achievement in System Administration. He was previously a performance lead and kernel engineer at Sun Microsystems, where he developed the ZFS L2ARC and led performance investigations. He has also created numerous performance analysis tools, which have been included in multiple operating systems. His recent work includes developing methodologies and visualizations for performance analysis.

Long

Brendan Gregg is a senior performance architect at Netflix, where he does large scale computer performance design, analysis, and tuning. This work includes evaluating new technologies and opportunities, and provides performance expertise to guide major changes and projects. This can involve any application, at any level of the stack, and for cloud instances and physical servers.

At Netflix, Brendan has developed and used analysis tools for the JDK hotspot and Node.js v8 runtimes, the Linux and FreeBSD kernels, the Xen hypervisor, Linux containers, and Docker. He has expertise with and has made contributions to multiple technologies, including Linux perf, ftrace, enhanced BPF, PMCs, ZFS, and DTrace. He has developed, debugged, and tuned software in rings 0, 1, and 3.

Brendan also assists production triage for performance and availability issues, and has been the primary on-call engineer (on the CORE team rotation) to lead the worldwide response for Netflix outages.

He is the author of the book "Systems Performance: Enterprise and the Cloud" (Prentice Hall, 2013), the primary author of "DTrace: Dynamic Tracing in Oracle Solaris, Mac OS X, and FreeBSD" (Prentice Hall, 2011), and co-author of "Solaris Performance and Tools: DTrace and mdb techniques for Solaris 10 and OpenSolaris" (Prentice Hall, 2006), and well as numerous articles about systems performance. For contributions to the field of system administration, particularly his work in systems performance analysis methodologies, he is the recipient of the USENIX 2013 LISA Award for Outstanding Achievement in System Administration.

He was previously the lead performance engineer at Joyent, a San Francisco-based high performance cloud computing provider. There he analyzed performance and scalability for small- to large-sized cloud computing environments, at any level of the software stack, including hypervisor internals, down to metal. He also developed and taught classes in cloud performance and DTrace. Prior to that he was a performance lead and kernel engineer at Sun Microsystems and later Oracle, and led the performance efforts for the world's first ZFS-based storage appliance, helping make it the fastest product of its class. This work included performance analysis and tuning while working with Sun's performance and availability engineering group, as well as Solaris kernel engineering. This work improved the performance of the product and the Solaris kernel, specifically in areas including networking, storage, and observability. He has also previously worked as a performance consultant and instructor, and has taught classes worldwide including authoring and delivering his own workshops. Over 2,000 people from across the industry have been taught by Brendan.

Brendan has invented and developed numerous advanced performance analysis tools, methodologies, and visualizations. Many of his tools, especially DTrace scripts, are shipped by default in Mac OS X, FreeBSD, Oracle Solaris 11, and are used by other performance analysis products. He created and developed the DTraceToolkit, a collection of these performance analysis tools, and more recently the perf-tools collection, of similar tools for the existing tracing frameworks on Linux. Apart from observability tools, he developed the ZFS L2ARC – a pioneering performance technology which uses flash memory as a second level cache, after main memory.

His recent work has included new performance methodologies and visualizations for Linux kernel analysis, and performance tools for both Linux and FreeBSD. His methodologies include the USE method and the TSA method, and his visualizations include latency, utilization, and sub-second offset heat maps, and flame graphs.

Born in Australia and later working in the Asia Pacific region, he has lived in the US since 2006.

(In case my bio inspires other Australians or non-US citizens to consider working in the US, then I'd recommend that you get a green card beforehand, else consider remote work instead of a more restrictive visa.)

For selected examples of my work, see my portfolio page.

Interviews

Talks

Upcoming talks:

Conferences I will or have spoken at. Mouse-over for names (tooltip), click for websites:

Prior conference talks include (see my homepage for slides and video):

  1. "Linux 4.x Tracing Tools: Using BPF Superpowers", Brendan Gregg, USENIX/LISA, Boston 2016
  2. "Designing Tracing Tools", Brendan Gregg, Sysdig CCWFS, San Francisco 2016
  3. "Java Performance Analysis on Linux with Flame Graphs", Brendan Gregg, JavaOne, San Francisco 2016
  4. "System Methodology - Holistic Performance Analysis on Modern Systems" (keynote), Brendan Gregg, ACM Applicative, New York 2016
  5. "Linux Systems Performance", Brendan Gregg, PerconaLive, Santa Clara 2016
  6. "Performance Checklists for SREs", Brendan Gregg, USENIX SREcon, Santa Clara 2016
  7. "Linux 4.x Performance: Using BPF Superpowers", Brendan Gregg, Performance@Scale, Menlo Park 2016
  8. "Broken Linux Performance Tools", Brendan Gregg, Southern California Linux Expo (SCALE14x), Los Angeles 2016
  9. "Broken Performance Tools", Brendan Gregg, QCon, San Francisco 2015
  10. "Java Mixed-Mode Flame Graphs", Brendan Gregg, JavaOne, San Francisco 2015
  11. "Netflix Instance Performance Analysis Requirements", Brendan Gregg, Monitorama, Portland 2015
  12. "Linux Profiling at Netflix", Brendan Gregg, Southern California Linux Expo (SCALE13x), Los Angeles 2015
  13. "Linux Performance Analysis: New Tools and Old Secrets", Large Installation System Administration (LISA/USENIX), Seattle 2014
  14. "Performance Tuning EC2 Instances", Brendan Gregg, AWS re:Invent, Las Vegas 2014
  15. "FreeBSD Flame Graphs", Brendan Gregg, FreeBSD Developer and Vendor Summit, San Jose 2014
  16. "Performance Analysis", Brendan Gregg, MeetBSD California, San Jose 2014
  17. "Linux Performance Tools", Brendan Gregg, LinuxCon | CloudOpen Europe, Düsseldorf 2014
  18. "From DTrace to Linux", Brendan Gregg, Tracing Summit, Düsseldorf 2014
  19. "From Clouds to Roots", Brendan Gregg, SurgeCon, Baltimore 2014
  20. "Linux Performance Tools", Brendan Gregg, LinuxCon | CloudOpen North America, Chicago 2014
  21. "Analyzing OS X Systems Performance with the USE Method", Brendan Gregg, MacIT, San Francisco 2014
  22. "What Linux can learn from Solaris performance and vice-versa" (keynote), Brendan Gregg, SCaLE12x, Los Angeles 2014
  23. "Blazing Performance with Flame Graphs" (plenary), Brendan Gregg, Large Installation System Admin. (LISA/USENIX), Washington DC, 2013
  24. "Open Source Systems Performance", Brendan Gregg, O'Reilly Open Source Convention (OSCON), Portland 2013
  25. "Stop the Guessing: Performance Methodologies for Production Systems", Brendan Gregg, Velocity, Santa Clara 2013
  26. "Linux Performance Analysis and Tools", Brendan Gregg, Southern California Linux Expo (SCaLE11x), Los Angeles 2013
  27. "Performance Analysis Methodology", Brendan Gregg, Large Installation System Admin. (LISA/USENIX), San Diego 2012
  28. "ZFS Performance Analysis and Tools", Brendan Gregg, zfsday, San Francisco 2012
  29. "DTracing the Cloud", Brendan Gregg, illumosday, San Francisco 2012
  30. "Real-time in the real world: DIRT in production", Bryan Cantrill, Brendan Gregg, Surge, Baltimore 2012
  31. "The USE Method", Brendan Gregg, Fórum Internacional Software Livre 13 (FISL13), Porto Alegre 2012
  32. "Visualizations", Brendan Gregg, dtrace.conf, San Francisco 2012
  33. "Performance Analysis: new tools and concepts from the cloud", Brendan Gregg, SCaLE10x, Los Angeles 2012
  34. "Design and Implementation of a Real-Time Cloud Analytics Platform", Dave Pacheco, Brendan Gregg, OSCON data, Portland 2011
  35. "Breaking Down Query Latency With DTrace", Brendan Gregg, Percona Live, New York 2011
  36. "Visualizations for Performance Analysis", Brendan Gregg, Large Installation System Admininistration (USENIX/LISA), San Jose 2010
  37. "How To Build Better Applications With Oracle Solaris DTrace", Brendan Gregg, Jim Mauro, Oracle Open World (OOW), San Francisco 2010
  38. "DTrace Crash Course", Brendan Gregg, Kernel Conference Australia (KCA), Brisbane 2009
  39. "Kernel Engineering", Brendan Gregg, Kernel Conference Australia (KCA), Brisbane 2009

Prior tutorials/workshops:

  1. "Linux Performance Analysis Using Perf and BPF", Brendan Gregg, Sasha Goldshtein. USENIX/LISA, Boston 2016
  2. "Linux Performance Tools", Brendan Gregg, O'Reilly Velocity, Santa Clara 2015
  3. "Metrics Workshop", Brendan Gregg, Narayan Desai, Kent Skaar, Theo Schlossnagle, Caskey Dickson. Large Installation System Admin (LISA/USENIX), Washington DC 2013

Not listed here are instructor led training classes I've developed and delivered, which ranged from 3 to 10 days in length.


Last updated: 30-Dec-2016