Master Linux Disk Management: Devices, RAID, Partitioning, and LVM

Disk Management

Linux treats everything as a file, including hardware devices that appear as device files.

# ll /dev/sda*
brw-rw---- 1 root disk 8, 0 Nov 20 04:11 /dev/sda
brw-rw---- 1 root disk 8, 1 Nov 20 04:11 /dev/sda1
brw-rw---- 1 root disk 8, 2 Nov 20 04:11 /dev/sda2
# ll /dev/zero
crw-rw-rw- 1 root root 1, 5 Nov 20 04:11 /dev/zero

Major number identifies device type; minor number identifies a specific device.

1. Disk External Structure

Disk classifications:

Solid‑state drive (SSD): internal architecture similar to a USB flash drive.

Mechanical hard drive (HDD): rotating platters, spindle, actuator arm, similar to a DVD.

NVMe drive.

PCI‑E interface.

Size categories:

3.5‑inch – desktop computers.

2.5‑inch – servers and laptops.

Interface types:

IDE – obsolete.

SCSI – obsolete.

SATA – desktop and laptop.

SAS – enterprise server standard.

SSD is faster but more expensive and has limited lifespan; HDD is cheaper, larger, and slower. Performance improves with higher rotation speed (5400, 7200, 10K, 15K rpm). Note: speed is not solely determined by interface; NVMe drives are the fastest.

2. Disk Array (RAID)

Purpose:
  Obtain larger capacity   # combine multiple disks into one
  Increase performance    # write to two disks faster than one
  Improve reliability     # mirror data on two disks

Interview questions

RAID levels summary:

RAID 0 – at least one disk, usable capacity = total size, no safety, fastest read/write, suitable for speed‑critical workloads like database replicas.

RAID 1 – exactly two disks, usable capacity = half, can lose one disk, slower write, typical for system disks.

RAID 5 – at least three disks, usable capacity = n‑1 disks, can lose one disk, balanced performance, suitable for stable business traffic.

RAID 10 – at least four disks (multiple of two), usable capacity = half, can lose half the disks, high read/write speed, ideal for high‑concurrency scenarios.

3. Disk Partitioning

Windows default uses MBR, which supports up to four primary partitions (C D E F). MBR can also have three primary partitions plus one extended partition containing many logical partitions.

Linux device naming example:

sda   # first disk
sda1  # first partition of first disk
sda2  # second partition of first disk
sdb   # second disk
sdb1  # first partition of second disk
sdb5  # first logical partition of second disk

1. Linux partition schemes

# System partition example (300 GB disk)
/boot 200M   # kernel and bootloader
/      rest  # root filesystem

# Swap partition example
/boot 200M
swap 2G   # temporary memory when RAM is insufficient

# Custom example
/boot 200M
swap 2G
/ 50G   # system
/data 1.8T   # data partition

2. Partition types

MBR: up to 4 primary partitions. GPT: up to 128 primary partitions.

3. Partition steps

1. For MBR (<2 TB) use fdisk.
2. For GPT (>2 TB) use parted.

# Scan SCSI bus
for i in $(ls /sys/class/scsi_host/); do echo '- - -' > /sys/class/scsi_host/${i}/scan; done

# fdisk example
fdisk /dev/sdb
# (use 'n' to create, 'p' for primary, specify size, etc.)

# partprobe to refresh kernel partition table
partprobe

Non‑interactive fdisk example:

# echo -e 'p
n

+1G
p
' | fdisk /dev/sdb

parted example (GPT):

(parted) mklabel gpt
(parted) mkpart primary xfs 0 500G
(parted) print

4. Filesystems

A filesystem defines how files are organized and stored on a storage device.

Common Linux filesystems:

ext2 – suitable for small, rarely‑updated partitions (e.g., /boot).

ext3 – ext2 with journaling.

ext4 – latest ext version with large file support, nanosecond timestamps, and performance improvements.

xfs – SGI, supports up to 8 EB.

swap – dedicated swap partition.

iso9660 – optical disc filesystem.

Common Windows filesystems:

FAT32 – max 4 GB file, 16 TB volume.

NTFS – max 16 EB file and volume.

exFAT.

Common Unix filesystems:

FFS (fast).

UFS – default on many Unix systems.

JF32.

3. Creating filesystems

# mkfs.ext4 /dev/sdb1
# mkfs.xfs /dev/sdb2

5. Disk Mounting

mount [-lhV]
mount -a [options]
mount --source /dev/sdb1 -o ro /dir1

Mount source can be a device file, label, UUID, or pseudo‑filesystem (proc, sysfs, devtmpfs, configfs). The mount point must exist and is usually an empty directory. Only one device can be mounted on a point at a time, but a device may be mounted on multiple points.

6. Persistent Mounts

Entries in /etc/fstab have six fields: device (or LABEL/UUID), mount point, filesystem type, mount options, dump frequency, and fsck order.

/dev/sdb1 /mnt/xfs xfs defaults 0 0

7. Common Disk Tools

df – display filesystem usage (e.g., df -t ext4 -h).

du – display directory usage (e.g., du -sh /etc/).

dd – low‑level copy, useful for backing up MBR ( dd if=/dev/sdb of=./sdb.img bs=512 count=1) or entire disks.

8. Enterprise Cases

Case 1: Swap sizing and creation

# dd if=/dev/zero of=/swapfile bs=1G count=1
# mkswap /swapfile
# swapon /swapfile

Case 2: Disk full – locate large files # find / -type f -size +1G Case 3: Insufficient space – add larger disk, mount, move log, create symlink

# mount /dev/sdc3 /data
# mv /var/log/10g /data/
# ln -s /data/10g /var/log/10g

Case 4: Deleting a file still held by a process

# lsof | grep 10g
# kill -9 <pid>

9. Logical Volumes (LVM)

LVM provides an abstraction layer for flexible storage management: physical volumes → volume group → logical volume → filesystem.

1. What is a logical volume

LVM allows resizing, adding, and moving logical volumes across multiple physical devices.

2. Creating logical volumes

# pvcreate /dev/sdb1 /dev/sdb2 /dev/sdb3
# vgcreate vg1 /dev/sdb1 /dev/sdb2
# lvcreate -L 5G -n lv1 vg1
# mkfs.ext4 /dev/vg1/lv1
# mount /dev/vg1/lv1 /lv1

3. Extending logical volumes

# vgextend vg1 /dev/sdb3
# lvextend -L 7G /dev/vg1/lv1
# resize2fs /dev/vg1/lv1   # for ext* filesystems
# xfs_growfs /dev/vg1/lv2  # for XFS after lvextend