Storage & LVM

A block device (`/dev/sda`, `/dev/nvme0n1`) represents a raw disk as the kernel sees it — unusable for files directly until it's partitioned and formatted. Partitioning divides a disk into logical sections, recorded in a partition table using one of two schemes: the older MBR (Master Boot Record, limited to 4 primary partitions and 2TB disks) or the modern GPT (GUID Partition Table, supporting far larger disks and more partitions) — GPT is the standard choice for any new system today.

A filesystem then formats a partition (or an LVM logical volume) with a structure for actually organizing files and directories — ext4 is the long-standing, reliable default on most Linux distributions; XFS is common on RHEL-family systems and excels with very large files and filesystems; Btrfs offers built-in snapshots and checksumming at the cost of more complexity. Creating a filesystem (`mkfs`) is what turns an empty partition into something `mount` can actually attach to the directory tree.

Mounting attaches a filesystem to a specific point in the existing directory tree, making its contents accessible at that path — `mount /dev/sdb1 /mnt/data` makes that partition's files appear under `/mnt/data`. A mount performed manually like this doesn't survive a reboot; `/etc/fstab` is the file that defines mounts to be applied automatically at boot, referencing each filesystem by its UUID (a stable identifier, preferred over a device name like `/dev/sdb1` which can shift if disks are added or removed) rather than a device path that isn't guaranteed stable across reboots.

Logical Volume Management (LVM) adds a flexible layer between raw disks and filesystems, solving a real limitation of plain partitions: a plain partition's size is essentially fixed at creation and awkward to resize, especially to grow across what used to be free space on a different disk entirely. LVM's architecture has three layers: a Physical Volume (PV) is a disk or partition initialized for LVM use; a Volume Group (VG) pools one or more PVs together into a single unit of available storage, regardless of how many physical disks back it; a Logical Volume (LV) is then carved out of a VG's pooled space, and it's the LV that actually gets formatted with a filesystem and mounted — behaving, from the filesystem's perspective, just like a regular partition would.

The payoff is flexibility: you can extend a Volume Group by simply adding another Physical Volume (another disk) to it, and then extend a Logical Volume to use that newly available space — growing storage without needing to have predicted the exact right size upfront, and often without unmounting anything (ext4 and XFS both support online, live resizing when growing, though shrinking is more restrictive and filesystem-dependent).

The complete build chain: `pvcreate` initializes a disk/partition as a PV. `vgcreate` creates a VG from one or more PVs. `lvcreate` carves out an LV from the VG's free space. `mkfs` formats the LV with a filesystem exactly as you would a plain partition. `mount` attaches it. To later grow it: `lvextend` grows the LV's size within the VG's available space, and then a filesystem-specific resize command (`resize2fs` for ext4, `xfs_growfs` for XFS) extends the filesystem itself to actually use that newly available space — extending the LV alone does not automatically grow the filesystem sitting on top of it.