Disk partition

A disk partition is a contiguous range of logical blocks on a storage device that is recorded in a partition table and treated by the operating system as an independent unit. In OS storage, partitions are the layer between the raw disk and the file system: each one can hold its own file system, boot loader or recovery environment, and each can be formatted, encrypted or erased without touching the others.

Partitioning is how operating systems separate system files from user data, keep firmware boot code in a known place and allow several operating systems to share one physical drive. The concept looks simple, but its details, such as the partition table type, alignment and size limits, decide whether a disk boots at all.

How a disk partition is defined

A disk partition is defined by a start address and a length in logical blocks, stored as one entry in the disk’s partition table. The operating system reads that table at startup and presents each partition as a separate block device, such as a volume letter on Windows or a node like /dev/sda2 or disk0s2 on Linux and macOS.

Modern drives expose logical blocks of 512 bytes or 4,096 bytes, addressed by logical block addressing (LBA). The partition entry also carries a type identifier that tells software what the partition is for, for example an EFI system partition, a Windows data partition or a Linux swap area. A partition entry does not contain a file system; it only reserves space that formatting later fills.

Two partition table formats dominate desktop computing, and the choice between them sets hard limits.

Because GPT keeps a backup copy at the end of the disk and protects headers with CRC32 checksums, a damaged primary table can usually be rebuilt. An MBR disk has no such copy, which makes its single sector a critical point of failure.

Common partition types and their roles

A typical system disk contains several partitions with fixed roles, and only one of them is the volume that users see. The others support booting, recovery and firmware updates, and deleting them can leave a computer unable to start.

Apple systems handle this layer differently. An APFS container occupies one GPT partition and divides its space dynamically among volumes, so macOS rarely needs additional partitions. In Windows and Linux, by contrast, each partition has a fixed size until it is explicitly resized.

Partitions, volumes and file systems

A partition, a volume and a file system are three different layers that are often confused. The partition is a reserved range of blocks, the file system is the structure that organizes files inside that range, and the volume is the mounted, usable result that the operating system presents.

The distinction matters during troubleshooting. A disk can have a valid partition with a corrupted file system, which shows up as a RAW or unreadable volume, while the partition table itself is fine. Repair tools for each layer are different, and using the wrong one can make recovery harder.

Planning and resizing partitions

Good partition planning keeps the layout as simple as the use case allows. On a single-user laptop, one system partition plus the required boot and recovery partitions is usually best, because fixed partitions waste free space that another partition might need later.

• Use GPT on every disk larger than 2 TiB and on any UEFI system disk.
• Keep partitions aligned to 1 MiB boundaries, which every current tool does by default and which prevents slow read-modify-write cycles on SSDs and Advanced Format drives.
• Create a separate data partition only when there is a clear reason, such as dual boot or a planned clean install of the system partition.
• Shrink a volume only after a backup, because moving data during a resize can fail on power loss.
• Leave recovery and EFI partitions in place unless the boot configuration is fully understood.

Resizing works from the end of a partition more easily than from the start. Extending a partition requires unallocated space directly after it, which is why a recovery partition placed behind the system partition often blocks Windows Disk Management from growing drive C.

Risks and common misconceptions

The most widespread misconception is that partitioning protects data. A second partition on the same drive fails together with it, so a partition is never a substitute for a backup on a separate device.

Deleting a partition removes only its table entry, so the data often remains recoverable until it is overwritten, which matters both for accidental deletion and for secure disposal. Converting between MBR and GPT is possible without data loss using tools such as mbr2gpt on Windows, but any table operation should follow a verified backup.

The simple explanation

A disk partition is like a fenced section of a large field. The field is the whole drive, and each fence marks off an area that can be used for something different, such as starting the computer, storing the system or keeping a repair kit.

The fences do not move by themselves, and a fire in the field affects every section. Partitions organize a disk, but they do not keep data safe if the drive itself fails.

How it is used in practice

A software developer buys a 2 TB NVMe drive for a desktop that will run both Windows and Linux. During setup, the drive is initialized with GPT, and the Windows installer creates the EFI system partition, the reserved partition, a 900 GB Windows partition and a recovery partition.

The Linux installer then uses the remaining space for an ext4 root partition and reuses the existing EFI system partition for its boot loader, so the firmware boot menu offers both systems. Project files are kept on a separate network share rather than on a shared partition.

A year later the Windows partition fills up. Because the Linux partition sits directly behind it, the developer backs up both systems first, then shrinks the Linux partition from its start with a bootable partition editor and extends Windows. The work takes about an hour and both systems boot normally afterward.