Linux storage is a layered stack: physical or virtual block devices sit at the hardware layer, volume managers (LVM, mdadm) and encryption (LUKS2) sit in the middle, and filesystems (Ext4, XFS, Btrfs, OpenZFS) or network/distributed protocols (Ceph, NFS/SMB, iSCSI/NVMe-oF) sit on top.
Linux filesystems
Extent trees, flex_bg, journaling modes (data=ordered, journal, writeback), fast_commit, tune2fs, and online resizing.
Allocation Groups (AGs), multi-threaded allocation, reflink CoW deduplication, dynamic inodes, and RAID stripe alignment.
CoW architecture, subvolumes, snapshots, send/receive, zstd compression, scrub self-healing, and RAID profiles.
Pool creation, RAIDZ, datasets, zvols, snapshots, send/receive, scrub, compression, dedup, ARC/L2ARC/SLOG, and encryption.
Storage security & encryption
Argon2id KDF, 32 keyslots, header backups, NVMe workqueue bypass flags, crypttab, and LVM-on-LUKS.
Automated boot unlocking: TPM2 PCR binding via systemd-cryptenroll, Network-Bound Disk Encryption (Tang), and FIDO2 keys.
Kernel I/O & performance tuning
Multi-queue block layer, none vs mq-deadline vs bfq vs kyber, sysfs queue parameters, and persistent udev rules.
Eliminating I/O freezes: vm.dirty_ratio, vm.dirty_background_ratio, vfs_cache_pressure, O_DIRECT, and database profiles.
Submission and Completion Queue ring buffers, zero-syscall I/O, SQPOLL kernel threads, fixed buffers, and fio benchmarks.
fio, dd, iostat, iotop, hdparm, biolatency, ioping. Measuring IOPS, bandwidth, latency distributions, and thermal limits.
fstrim, discard mount option, blkdiscard, NVMe format, RAID TRIM, dm-crypt, LVM thin pools, and Btrfs async discard.
bcache, lvmcache, dm-cache, and dm-writecache. Writethrough vs writeback, setup, performance tuning, and decision matrices.
Volume management & redundancy
Physical volumes, volume groups, logical volumes, snapshots, thin provisioning, LVM RAID, caching, and metadata recovery.
RAID 0/1/5/6/10, array creation, monitoring, disk replacement, growing arrays, write hole, bitmaps, and superblock versions.
The 3-2-1 rule, RPO/RTO, rsync, BorgBackup, Restic, database backups, snapshot-based backups, encryption, and DR planning.
e2fsck for ext4, xfs_repair for XFS, btrfs check, superblock recovery, journal replay, bad blocks, and corruption prevention.
Physical, cloud & distributed storage
Namespaces, 4Kn native vs 512e LBA formatting, Host Memory Buffer (HMB), power states, and nvme-cli commands.
Online disk expansion on AWS EBS, GCP Persistent Disk, and Azure Disks: kernel rescan, growpart, LVM, and cloud-init.
RADOS architecture, CRUSH data placement, mapping kernel RBD block devices, CephFS mounts, and BlueStore.
NFS v3/v4/v4.1/v4.2, Samba/SMB, exports, security (Kerberos, WireGuard), performance tuning, and Active Directory.
Block-level network storage: iSCSI targets with targetcli, NVMe over Fabrics (TCP/RDMA/FC), initiators, and CHAP.
OverlayFS layers, whiteouts, copy-up, Docker/Podman storage drivers, volumes, bind mounts, and Kubernetes PV/PVC.
Redundant paths to SAN storage, multipathd, /etc/multipath.conf, path grouping, failover, and LVM integration.
User, group, and project quotas on ext4, XFS, Btrfs, and ZFS. Soft vs hard limits, grace periods, and NFS quotas.
SRE incident troubleshooting runbooks
Emergency production runbooks: 100% full disk with unlinked deleted files (lsof +L1), emergency read-only remounts, target is busy unmount errors, and degraded RAID arrays.
Frequently asked questions
What topics does the Linux storage section cover?
Linux filesystems (Ext4, XFS, Btrfs, OpenZFS), volume management (LVM, mdadm RAID), block encryption (LUKS2, TPM2, Clevis/Tang), kernel I/O (blk-mq schedulers, page cache, io_uring), physical & distributed storage (NVMe, Ceph, Cloud EBS/PD, NFS, iSCSI, NVMe-oF), and SRE troubleshooting runbooks.
Which Linux filesystems are documented in depth?
Ext4, XFS, Btrfs, and OpenZFS are covered in comprehensive dedicated guides, detailing internal on-disk structures, allocation groups, copy-on-write, extents, journaling modes, compression, and online resizing.
What encryption mechanisms are covered?
LUKS2 with dm-crypt and Argon2id KDF, performance workqueue optimization for NVMe, and automated unattended boot unlock using TPM 2.0 PCR policies and Clevis/Tang Network-Bound Disk Encryption (NBDE).
How do you choose between Linux I/O schedulers?
Use none for NVMe SSDs with hardware queues, mq-deadline for SATA/SAS SSDs and database workloads, bfq for rotational spinning HDDs, and kyber for latency-targeted cloud services.
What distributed and cloud storage systems are documented?
Ceph (RADOS, RBD kernel block mapping, CephFS), cloud block storage online expansion (AWS EBS, GCP Persistent Disk, Azure Disks with growpart and cloud-init), and enterprise SAN protocols (iSCSI, NVMe-oF, DM-Multipath).