エピソード

  • How Linux Handles Memory Fragmentation on Production Servers

    How Linux Handles Memory Fragmentation on Production Servers
    2026/06/07
    Episode 36 of The Linux Podcast dives into memory fragmentation, a silent performance killer on long-running Linux servers. Lucas explains how the kernel's buddy allocator causes external fragmentation over time, and how newer features like THP compaction, proactive compaction via khugepaged, and the 'compact_memory' sysfs interface help. Luna shares a real-world case where a Redis instance on a 128GB machine saw 40% tail latency spikes due to fragmentation. They discuss practical monitoring with /proc/pagetypeinfo, when to trigger compaction manually, and why some workloads benefit from transparent huge pages while others should disable them. No theory without application: listeners will learn one command to check their server's fragmentation level today. #Linux #MemoryManagement #Fragmentation #Kernel #BuddyAllocator #TransparentHugePages #THP #ProactiveCompaction #khugepaged #Redis #SysAdmin #Performance #ServerOptimization #Technology #FexingoBusiness #BusinessPodcast #OpenSource #Systems Keep every episode free: buymeacoffee.com/fexingo
    続きを読む 一部表示
    8 分
  • Why Linux Distros Are Adopting Composefs for Container Storage

    Why Linux Distros Are Adopting Composefs for Container Storage
    2026/06/06
    Container images are huge, duplicative, and a security headache. Lucas and Luna dig into Composefs — a new Linux filesystem layer developed by Red Hat's Alexander Larsson — that uses content-addressable storage to deduplicate image data across the entire host. They walk through how it works, why it slashes disk usage by up to 90 percent for common image stacks, and what it means for container runtime performance and image signing. With OCI image sizes ballooning and supply chain attacks on the rise, Composefs offers a kernel-level fix that's landing in Fedora 40 and beyond. This episode covers the technical meat without assuming you're a kernel developer. #Composefs #LinuxContainers #Podman #OCI #RedHat #ContainerStorage #AlexanderLarsson #DeviceMapper #OverlayFS #ImageDeduplication #Fedora40 #LinuxKernel #SupplyChainSecurity #CamelCase #Technology #LinuxPodcast #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo
    続きを読む 一部表示
    11 分
  • Why Linux Memory Tiering Is Changing How Servers Use RAM

    Why Linux Memory Tiering Is Changing How Servers Use RAM
    2026/06/06
    Episode 34 dives into Linux memory tiering — a kernel feature that treats fast and slow memory as a single pool with automatic hot-page migration. Lucas explains how the kernel's memory-management subsystem can now move frequently accessed pages to high-bandwidth memory (like Intel Optane or CXL-attached RAM) while relegating cold data to cheaper DRAM tiers. Using concrete examples from Meta's production servers and the Linux 6.8 kernel merge, the hosts discuss how this changes database performance, cloud instance pricing, and the old NUMA-awareness playbook. Luna asks whether this makes manual memory pinning obsolete, and Lucas walks through the trade-offs: latency jitter, kernel overhead, and the ongoing debate about transparent vs. user-hinted migration. No hype — just how the kernel is solving a hardware problem that didn't exist five years ago. #Linux #MemoryTiering #Kernel #NUMA #CXL #IntelOptane #Meta #ProductionServers #RAM #HotPages #MemoryManagement #LinuxKernel68 #DatabasePerformance #CloudComputing #Technology #OpenSource #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo
    続きを読む 一部表示
    10 分
  • Why Linux Is Taking Over Automotive Infotainment

    Why Linux Is Taking Over Automotive Infotainment
    2026/06/05
    Episode 33 of The Linux Podcast explores why carmakers from Toyota to Mercedes are ditching proprietary OSes for Linux-based infotainment systems. Lucas breaks down the Automotive Grade Linux (AGL) consortium, how the shared platform cuts development costs by 40%, and why real-time kernel patches are essential for in-car safety. Luna asks whether drivers actually care what OS runs their dashboard, and the hosts debate the tension between open-source flexibility and automaker control. A concrete look at how open source is quietly steering the future of your car's dashboard. #Linux #AutomotiveGradeLinux #AGL #Infotainment #OpenSource #Automotive #RealTimeKernel #LinuxFoundation #Toyota #Mercedes #Technology #SoftwareDefinedVehicle #Genivi #YoctoProject #LinuxKernel #FexingoTechnology #Fexingo #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo
    続きを読む 一部表示
    8 分
  • Why Linux AI Servers Need Real-Time Kernels Now

    Why Linux AI Servers Need Real-Time Kernels Now
    2026/06/05
    Episode 32 of The Linux Podcast with Fexingo dives into a growing tension in the Linux ecosystem: AI inference at the edge and in data centers demands deterministic latency, but standard Linux kernels prioritize throughput over real-time guarantees. Lucas and Luna explore why the PREEMPT_RT patch set, merged into the mainline kernel in 2024, is suddenly getting serious attention from NVIDIA, Canonical, and Red Hat. They break down a concrete example: a self-driving car stack running on an NVIDIA Orin system-on-chip, where a jitter spike of just 10 milliseconds can mean a missed sensor fusion deadline. The episode explains how real-time Linux works under the hood, why the audio and industrial automation worlds have used it for years, and what changes when AI inference meets hard deadlines. No hype, just the architecture — and why this matters for anyone building Linux-based AI systems in 2026. #Linux #RealTimeLinux #PREEMPT_RT #AI #EdgeInference #NVIDIA #Canonical #RedHat #Orin #SelfDrivingCars #Kernel #Latency #Jitter #Technology #OpenSource #FexingoBusiness #BusinessPodcast #TechPodcast Keep every episode free: buymeacoffee.com/fexingo
    続きを読む 一部表示
    8 分
  • BPF Is the Linux Superpower You Never Knew You Had

    BPF Is the Linux Superpower You Never Knew You Had
    2026/06/04
    Episode 31 of The Linux Podcast with Fexingo dives into BPF, the Linux kernel's secret weapon for observability, security, and performance tuning. Lucas and Luna break down how BPF—originally the Berkeley Packet Filter—has evolved into a virtual machine inside the kernel, letting administrators run sandboxed programs to trace system calls, monitor network packets, and even patch running kernels without rebooting. They walk through real-world use cases: how Netflix uses BPF for continuous profiling, how Cloudflare leverages XDP for DDoS mitigation, and why BPF is replacing traditional monitoring tools like strace and tcpdump. They also discuss the learning curve and the emerging ecosystem of front-end tools like bpftrace and Cilium. If you manage Linux servers or just want to understand what makes modern Linux tick, this episode gives you the concrete hook to start exploring BPF today. #BPF #eBPF #LinuxKernel #Observability #Netflix #Cloudflare #XDP #bpftrace #Cilium #LinuxPerformance #KernelTracing #OpenSource #DevOps #SysAdmin #Technology #FexingoBusiness #BusinessPodcast #LinuxPodcast Keep every episode free: buymeacoffee.com/fexingo
    続きを読む 一部表示
    8 分
  • Why Linux Container Runtimes Are Splitting in Two

    Why Linux Container Runtimes Are Splitting in Two
    2026/06/04
    Episode 30 of The Linux Podcast with Fexingo: Open Source Operating Systems, Distros, and Server Stack. Lucas and Luna dig into the quiet but consequential fragmentation of Linux container runtimes. For years, Docker's containerd and Red Hat's CRI-O served as the two main options, but a new wave of lightweight, purpose-built runtimes like youki, crun, and gVisor are challenging the status quo. The hosts trace how the shift from monolithic daemons to modular OCI-compliant tools began with Kubernetes dropping Docker, and explain why the runtime layer is now splitting along two axes: performance-sensitive vs. security-isolated. They cite specific benchmarks showing youki starting containers in under 50 milliseconds compared to containerd's ~200ms, and discuss why hyperscalers like Google and AWS are investing in gVisor and Firecracker. The episode closes with a forward look at what this means for the average sysadmin and whether the fragmentation will lead to innovation or just more complexity. #Linux #ContainerRuntimes #Containerd #CRIO #Youki #Crun #GVisor #Firecracker #OCI #Kubernetes #Docker #RedHat #Google #AWS #Sysadmin #Technology #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo
    続きを読む 一部表示
    8 分
  • How Linux Distros Are Securing the Supply Chain with SBOMs

    How Linux Distros Are Securing the Supply Chain with SBOMs
    2026/06/03
    Supply chain attacks on open source software are surging, and Linux distributions are fighting back with a tool called the software bill of materials, or SBOM. In this episode, Lucas and Luna break down how distros like Fedora and Alpine are adopting SBOMs to provide a transparent list of every dependency in a package. They discuss a real attack on the xz-utils library in 2024 that bypassed maintainer scrutiny for years, and explain how SBOMs could have caught it earlier. The conversation covers the tension between SBOM completeness and developer usability, why container images make the problem harder, and the role of tools like SPDX and CycloneDX in standardizing the format. If you use Linux on a server, in a container, or on the desktop, your security posture depends on knowing what's actually in your software stack. #Linux #OpenSource #SBOM #SupplyChainSecurity #SoftwareBillOfMaterials #CycloneDX #SPDX #Fedora #AlpineLinux #xzUtils #ContainerSecurity #DependencyManagement #DevOps #Security #Technology #FexingoBusiness #BusinessPodcast #TechPodcast Keep every episode free: buymeacoffee.com/fexingo
    続きを読む 一部表示
    13 分