In this episode, Lucas and Luna dive into the evolution of the Linux LIO (Linux-IO) target framework and its role in enabling NVMe-over-Fabrics in enterprise storage environments. They explore how LIO, originally built for SCSI, now handles NVMe-oF traffic, the impact on end-to-end NVMe ecosystems, and what this means for data center performance and protocol coexistence. Specific focus on LIO's 'backstore' abstraction and how it allows NVMe namespaces to be exported over Fibre Channel without dedicated hardware. Features a concrete discussion on the practical hurdles of transitioning from FCP to FC-NVMe, including fabric zoning, initiator compatibility, and multipathing.

#Linux #LIO #NVMe-oF #NVMe #FibreChannel #FC-NVMe #Storage #Enterprise #SCSI #Backstore #TargetDriver #DataCenter #OpenSource #Kernel #Technology #FexingoBusiness #BusinessPodcast #LinuxPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode of The Linux Podcast, Lucas and Luna explore how combining NFS with Remote Direct Memory Access (RDMA) is rewriting the rules for high-performance storage in data centers. They break down the technical mechanics — zero-copy data transfers, bypassing the kernel network stack — and explain why this matters for workloads like AI training and database clusters. The discussion is anchored on a real-world case: Mellanox's 2019 benchmarks showing 40% lower latency and 60% higher throughput for NFS over RDMA compared to traditional TCP. They also cover the Linux kernel's NFSoRDMA client support (added in kernel 4.8) and how it fits into the broader RDMA ecosystem (InfiniBand vs. RoCE v2). The hosts also consider tradeoffs: the need for specialized hardware, the complexity of tuning, and when standard NFS over TCP still makes sense. If you're a sysadmin or architect grappling with storage bottlenecks, this episode gives you a concrete tool to consider.

#Linux #NFS #RDMA #NFSoRDMA #Storage #DataCenter #Kernel #InfiniBand #RoCE #Mellanox #ZeroCopy #NetworkStack #AIWorkloads #DatabasePerformance #LinuxStorage #FexingoBusiness #BusinessPodcast #Technology

Keep every episode free: buymeacoffee.com/fexingo

Episode 113 dives into Btrfs copy-on-write snapshots—a feature that makes near-instantaneous, space-efficient backups possible on Linux. Lucas breaks down how COW works under the hood, contrasting it with traditional LVM snapshots that can degrade performance. Luna asks why Btrfs isn't the default filesystem for more distributions despite these advantages, and they discuss real-world trade-offs: metadata checksumming, RAID56 quirks, and the famous Btrfs scrub command. The conversation covers a specific use case—rolling snapshots on a home server—and explains how the reflink capability deduplicates data at the filesystem level. They also touch on the current state of Btrfs in kernel 6.12 and where it fits alongside ZFS and ext4. If you manage Linux storage, this episode gives you one concrete technique you can implement today.

#Btrfs #CopyOnWrite #Linux #Snapshots #DataProtection #Filesystem #LinuxStorage #COW #Backups #Kernel6.12 #Reflink #Scrub #RAID56 #ZFS #Ext4 #HomeServer #OpenSource #FexingoBusiness

Keep every episode free: buymeacoffee.com/fexingo

Episode 112 of The Linux Podcast with Fexingo dives into the quiet revolution in Linux software RAID. Lucas and Luna break down how the md driver has evolved from a simple mirroring tool to a robust foundation for enterprise storage, supporting RAID 0, 1, 4, 5, 6, 10, and even nested levels. They focus on RAID 6's dual-parity resilience and the growing role of Linux RAID in hyperconverged infrastructure. With real-world examples like Ceph using md for local journal devices, they explain why Linux RAID remains relevant despite hardware RAID controllers and NVMe. The episode also covers recent kernel improvements: faster resync with the 'sync_max' tunable, better handling of write-intent bitmaps, and integration with dm-integrity for data checksums. Whether you're a sysadmin or a curious developer, you'll learn one concrete thing: how Linux RAID can outperform hardware RAID in cost and flexibility.

#Linux #RAID #Storage #EnterpriseLinux #Kernel #MdDriver #RAID6 #Ceph #Hyperconverged #Sysadmin #OpenSource #DataProtection #Technology #FexingoBusiness #BusinessPodcast #LinuxPodcast #ServerStorage #DiskArrays

Keep every episode free: buymeacoffee.com/fexingo

In this episode of The Linux Podcast, Lucas and Luna dive into the mechanics of Linux namespace isolation, the core building block of container security. They break down the six default namespaces—PID, mount, network, UTS, IPC, user—and explain how each one restricts a container's view of the system. Using a practical Docker example, they show how a containerized web server sees only its own process tree, filesystem, and network stack, while the host remains invisible. The conversation touches on the 2024 runc container breakout vulnerability (CVE-2024-3246) to highlight why namespace isolation alone isn't enough—and how user namespaces add an extra layer of privilege separation. Perfect for DevOps engineers, sysadmins, and anyone curious about what makes containers secure (or not).

#Linux #Namespace #ContainerSecurity #Docker #DevOps #SysAdmin #Kernel #CVE #UserNamespace #PIDNamespace #NetworkNamespace #MountNamespace #UTSNamespace #IPCNameSpace #Technology #FexingoBusiness #BusinessPodcast #ContainerBreakout

Keep every episode free: buymeacoffee.com/fexingo

In this episode, Lucas and Luna dive into Linux kernel live patching — the technology that lets system administrators apply critical security updates to a running kernel without a single reboot. They trace the evolution from the original kGraft and Kpatch projects to today's unified livepatch infrastructure in mainline Linux. Lucas walks through a concrete example: how a real-world Blue Screen of Death scenario in a major e-commerce deployment was avoided by live-patching a TCP vulnerability that could have cost millions in downtime. Luna digs into the technical constraints: what types of patches can be applied hot, why function boundaries matter, and why live patching still isn't a cure-all for kernel hang issues. They also touch on the ecosystem — from Canonical's Livepatch Service to Red Hat's Kpatch — and discuss why the technology is increasingly critical in the era of zero-day exploits and 24/7 operations. If you run servers and dread reboot windows, this episode might change how you think about kernel maintenance.

#Linux #LivePatching #Kernel #kGraft #Kpatch #Livepatch #Uptime #ServerMaintenance #Security #ZeroDay #Canonical #RedHat #BPF #LinuxKernel #SysAdmin #OpenSource #FexingoBusiness #TechnologyPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode of The Linux Podcast with Fexingo, Lucas and Luna dive into the mechanics of Linux Control Groups (cgroups v2), the kernel feature that underpins container resource management. They explore how cgroups limit CPU, memory, and I/O for processes, compare them to namespaces, and walk through a practical example of setting up a memory-bound cgroup on a server. With Docker and Kubernetes relying heavily on cgroups, understanding this technology is crucial for anyone running containerized workloads. The hosts also touch on the evolution from cgroups v1 to v2 and common pitfalls like the OOM killer interaction. Listeners will leave with a concrete grasp of how Linux keeps noisy neighbors in check.

#Linux #Cgroups #Containers #Kubernetes #Docker #ResourceManagement #CgroupsV2 #LinuxKernel #Namespaces #OOMKiller #SystemAdministration #DevOps #CloudNative #ProcessIsolation #Technology #LinuxPodcast #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In Episode 108 of The Linux Podcast with Fexingo, Lucas and Luna dive into Kernel Same-page Merging (KSM) — a memory-saving feature that deduplicates identical pages across virtual machines. They explain how KSM scans memory, merges pages, and shares them with copy-on-write semantics, all while balancing CPU overhead. The episode covers real-world impacts: a typical virtualized server running 10 identical VMs can save 30-50% memory using KSM. They discuss tuning parameters like pages_to_scan and sleep_millisecs, and highlight KSM's role in projects like QEMU/KVM and Kubernetes clusters. Lucas walks through a hypothetical deployment: 20 Ubuntu VMs with similar workloads, where KSM reduces memory from 80GB to under 50GB. Luna challenges the trade-offs — what about latency-sensitive apps? Lucas explains KSM's MADV_MERGEABLE hint and how to exclude critical processes. The episode ties to the broader push for efficiency in cloud data centers, where power and cooling costs drive innovation.

#Linux #KSM #MemoryManagement #Virtualization #Kernel #QEMU #KVM #Deduplication #CloudComputing #DataCenter #Tuning #OpenSource #Technology #Podcast #Fexingo #FexingoBusiness #BusinessPodcast #SystemsAdmin

Keep every episode free: buymeacoffee.com/fexingo

Episode 107 of The Linux Podcast. Lucas and Luna dive into Fanotify, the Linux kernel's file access monitoring framework. They explain how Fanotify differs from inotify, walk through its use of file descriptors and permission events, and highlight real-world applications in antivirus, backup, and system integrity tools. Specific focus on the difference between FAN_ACCESS and FAN_OPEN_PERM events, and how Fanotify's synchronous notification model enables proactive security responses. No prior episode covered file system event monitoring from this angle.

#Linux #Fanotify #Kernel #FileMonitoring #Security #OpenSource #Technology #Podcast #FexingoBusiness #BusinessPodcast #LinuxSecurity #Filesystem #Antivirus #Inotify #LWN #EricParis #Sysadmin #KernelDevelopment

Keep every episode free: buymeacoffee.com/fexingo

In this episode of The Linux Podcast, Lucas and Luna dive into the quiet revolution of BPF (Berkeley Packet Filter), now a core mechanism for extending the Linux kernel safely. They explore how BPF has evolved from a simple packet filter to a programmable infrastructure layer, replacing traditional kernel modules with sandboxed, verifiable programs. The discussion centers on a concrete case: how Netflix uses BPF to monitor millions of streaming sessions without kernel panics. Lucas explains BPF's verifier, which checks memory safety before execution, and contrasts it with the risks of kernel modules. Luna brings up the practical impact on cloud native environments, where BPF powers tools like Cilium and Falco. The episode covers why BPF is not a silver bullet, with trade-offs in complexity and debugging. A must-listen for system administrators and developers looking to understand the future of kernel extensibility.

#Linux #BPF #eBPF #Kernel #Netflix #Cilium #Falco #CloudNative #Observability #Performance #Security #OpenSource #LinuxKernel #Container #Kubernetes #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo