Performance testing talk from the field.
PerfBytes tracks performance engineering as it moves from old scripting-heavy habits into cloud, microservices, CI/CD, and developer-owned testing. The tone is technical, conversational, and loose, with working practitioners comparing scars, tools, and event-floor observations.
PRESS PLAY
PerfBytes is a long-running performance engineering show with a working-practitioner feel. It treats performance as a living discipline shaped by architecture, tooling, operations, and culture. The recent episodes spend less time on generic testing theory and more time on what performance engineers actually face now. Kubernetes gets discussed through pods, memory limits, CPU sharing, rollouts, and the uncomfortable question of what happens to users when a pod dies. Developer experience appears as another major thread. A guest explains how performance testing changes when developers no longer need to reverse engineer applications they built themselves. That leads naturally into tool choices, code-based testing, APIs, command lines, and automation that can live inside CI/CD. Observability is handled as a way to lower the work of diagnosis, not as a decorative dashboard layer. The job-market conversation looks at modern performance roles, cloud knowledge, data analysis, and machine learning patterns for comparing results across runs. PerfBytes also keeps one foot in the performance community itself. StarEast, WOPR, QSConf, and PAC episodes give the show a hallway-track quality, full of event plans, old colleagues, and field notes from people who still meet around performance work. The tone is informal and sometimes unruly. Spanish and French references, running jokes, and conference banter sit beside serious talk about response time, bottlenecks, cloud subscriptions, and test orchestration. The show works best when that mix lands: technical enough for specialists, human enough to show how the discipline is changing through actual conversations. It is not polished vendor theater. It is performance people arguing, laughing, remembering older tools, and trying to make sense of newer systems.
Made for: PerfBytes suits performance engineers, testers, developers taking on performance work, and technical leads responsible for reliable software behavior under load.
What sets it apart: The show blends detailed performance craft with community chatter from conferences and practitioners. It is comfortable moving from Kubernetes resource limits to job offers, tooling APIs, and old testing habits in the same conversational lane.
PerfBytes: Helping IT Professionals to improve Performance Practices!
Kubernetes kills your pod; Leandro and Henrik make sure you know who pulled the trigger.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A practical Kubernetes episode on pod limits, OOM kills, rollouts, and CPU sharing for performance engineers.
Kubernetes comes into focus as Leandro and Henrik work through the operational details that can make performance tests behave in surprising ways.
A practical look at moving performance testing closer to the developers who build the application.
Perfides frames performance testing as a question of ownership: what changes when developers take responsibility for measuring applications they already know from the inside?
A conference reunion frames response time as a practical question of usability, cost, and judgment.
A Star East 2024 reunion turns into a lively, grounded conversation about what performance really means. Marc and Leandro move from response time across disk input and output, web requests, application logins, and time spent at 100 percent CPU to the more serious question of…
A practical look at performance tooling across automation, CI/CD, APIs, reporting, and observability.
This Perfvides episode follows the performance engineer's toolkit as it moves beyond the old script-and-slam approach. The host and Henrik cover execution, reporting, analysis, automation, CI/CD triggers, APIs, command lines, and observability, keeping the focus on how these…
A grounded look at how performance testing is changing as AI enters the discussion.
Performance Job Offers looks at performance testing as a discipline in motion, starting with the hosts' plan for more episodes and their sense that the work still attracts interest.
Specialists walking through software's tradeoffs, one system at a time.
Founders and engineers on AI coding agents like Claude Code, and supply chain security.
The engineering decisions behind code, agents, and control.
Python tools, practical trade-offs, and the occasional programming joke.
Matt Stratton interviews DevOps practitioners on security, platform engineering, and trusting AI-written code.
PerfBytes turns a Kubernetes question into a performance-engineering primer, with Henrik defining orchestration, master nodes, worker nodes, and the scheduler.
Francesc Campoy Flores and Mathieu Leroux argue that performance work only moves to developers when tooling, incentives, and sponsorship all change together.
PerfBytes’ Mark Tomlinson and Leandro record a StarEast 2024 conversation about where AI fits into performance engineering.
["The PerfBytes episode \"Performance testing tools in 2024\" argues that a realistic performance plan starts with real traffic numbers, not a guessed load mix.", "Henrik and Leandro say they work for companies that build performance-related tools, then spend the episode trying to stay unbiased about Google Analytics, Adobe, Datadog, and other platforms.", "They compare access logs, observability backends, Sankey graphs, and query languages such as LogQL, PromQL, and DQL as ways to rebuild user journeys and diagnose bottlenecks.", "They also contrast Excel, Tableau, GitLab, Jenkins, GitHub Actions, WebPageTest, DevTools, and PageSpeed Insights with SLO/SLI-driven automation and synthetic checks across browser and protocol layers."]
Leandro and Henrik frame 'Performance Job Offers' around a shift from 'performance tester' roles to performance engineer roles that mention Kubernetes, CI/CD, observability, and scripting.
PerfBytes' host interviews Lisa Waugh, a developer experience testing engineer at IBM with 41 years at the company, about her path from 1990s DB2 batch testing into performance engineering and, more recently, cloud and container-based load testing.
Henrik and Leandro discuss how to start in performance testing in 2023, covering common questions and patterns.
Leandro and guest Federico Toledo discuss upcoming events in the performance testing world, including a workshop and a new conference.
The hosts discuss the upcoming physical and virtual Performance Advisory Council (PAC) events, including logistics and potential sponsors.