Adrian Cockcroft on performance engineering’s shift from kernels to AI
Adrian Cockcroft examines how performance engineering has moved from low-level kernel inspection toward workloads shaped by AI. The discussion is framed by P99 CONF’s five-year record of speakers questioning the p99 metric.
Adrian Cockcroft’s discussion centers on a shift in performance engineering: from inspecting low-level kernel behavior to understanding workloads influenced by AI. The summary ties this to P99 CONF’s five-year history, where speakers have repeatedly challenged the p99 metric. That context suggests the field is rethinking how performance is measured and optimized as software and cloud systems evolve. Because the available material is limited, broader claims about specific tools or outcomes would be speculative.
The shift may affect engineers, cloud operators, and users who depend on responsive services. If AI-shaped workloads become more common, performance practices could prioritize different signals than the p99 metric, potentially changing how teams diagnose bottlenecks and set service expectations. Society could feel this through reliability and speed of digital services, though the actual impact remains uncertain and depends on adoption.