Finding statistically significant changes in performance results has always been challenging but now that most of our code runs on hardware and infrastructure we don't own, we need methods and tools for detecting performance changes in noisy data. This talk will provide tips for understanding noise, where it comes, and techniques for fighting it. Plus some anecdotes and real-life examples from the world of open-source.
Interview:
What's the focus of your work these days?
I'm leading the development of Nyrkiö's performance analysis product.
What's the motivation for your talk at QCon London 2024?
I struggled with getting reliable benchmark results for a long time until I learned the tools and tricks discussed in my talk and I want to share these with everyone.
How would you describe your main persona and target audience for this session?
Senior engineers or architects.
Is there anything specific that you'd like people to walk away with after watching your session?
The knowledge of how to handle noise in benchmark results and the latest techniques for doing so.
Speaker
Matt Fleming
CTO @Nyrkiö, Former Linux Kernel Maintainer @Intel and @SUSE
Matt is the co-founder and CTO of Nyrkiö, a company that wants to arm every developer with the tools to understand the performance of their code. He has spent over 15 years working on low-level, high-performance systems and was previously the maintainer for the Linux kernel EFI subsystem while at Intel and SUSE. Most recently, Matt has focused on the performance of distributed systems and databases including leading a globally distributed team of engineers that contributed significant performance changes to the Apache Cassandra 5.0 release. He has co-authored papers on performance change detection and distributed systems testing and served on the ACM/SPEC ICPE program committee. Matt can often be found on Twitter, discussing topics such as software performance, benchmarking and statistics.
Find Matt Fleming at:
Date
Monday Apr 8 / 03:55PM BST ( 50 minutes )
Location
Windsor (5th Fl.)
