https://speakerdeck.com/hoyeonrhee 2026-05-26 10:38:01 -0400 This session walks through a real-world system performance investigation centered on an unexpected speedup observed during a container migration. What initially appeared to be a straightforward performance improvement gradually revealed itself to be extreme CPU thrashing on the original server. Through an iterative investigation process, the root cause was traced to CPU saturation and severe resource contention introduced by noisy applications sharing the same host. The talk also introduces the observability tools used throughout the analysis, ranging from standard counters such as uptime and perf stat to BPF-based dynamic tracing. This session walks through a real-world system performance investigation centered on an unexpected speedup observed during a container migration. What initially appeared to be a straightforward performance improvement gradually revealed itself to be extreme CPU thrashing on the original server. Through an iterative investigation process, the root cause was traced to CPU saturation and severe resource contention introduced by noisy applications sharing the same host. The talk also introduces the observability tools used throughout the analysis, ranging from standard counters such as uptime and perf stat to BPF-based dynamic tracing. Wed, 27 May 2026 00:00:00 -0400 https://speakerdeck.com/hoyeonrhee/system-performance-analysis-from-case-study-to-bpf-tool https://speakerdeck.com/hoyeonrhee/system-performance-analysis-from-case-study-to-bpf-tool This session dives into the core mechanisms of Linux kernel tracing. We explore dynamic instrumentation such as exception-based kprobe and code-patching ftrace, comparing internals and performance overhead. The talk also covers low-overhead, statically defined tracepoints and counter-based sampling via perf_event. Finally, we introduce BPF as a programmable execution layer that reuses and enhances tracing sources for flexible and efficient kernel observation. Presented at: SUSE Labs Conference 2026 This session dives into the core mechanisms of Linux kernel tracing. We explore dynamic instrumentation such as exception-based kprobe and code-patching ftrace, comparing internals and performance overhead. The talk also covers low-overhead, statically defined tracepoints and counter-based sampling via perf_event. Finally, we introduce BPF as a programmable execution layer that reuses and enhances tracing sources for flexible and efficient kernel observation. Presented at: SUSE Labs Conference 2026 Mon, 18 May 2026 00:00:00 -0400 https://speakerdeck.com/hoyeonrhee/linux-kernel-tracing-internals https://speakerdeck.com/hoyeonrhee/linux-kernel-tracing-internals