Every phone, laptop, USB stick, memory card, and server depends on a file system, yet most people never think about it until a transfer fails or a device shows up as “not readable.” A file system is the software layer that decides how data is broken into pieces, where those pieces live on a disk, and how the operating system finds them again. This article walks through the major file systems in use today, in plain language, with sourced facts on their limits, history, and typical uses. It is written for a global audience, since the same memory cards, phones, and hard drives move between homes and offices in Lagos, Nairobi, Manila, Mumbai, and everywhere in between.
What a file system actually does
A storage device such as a hard disk, SSD, or SD card is really just a huge number of tiny addressable slots capable of holding data. On their own, those slots have no concept of “files” or “folders.” A file system is the layer that imposes that structure. It keeps three things straight: which slots (usually called blocks, sectors, or clusters) belong to which file, what a file is named and where it sits in the folder hierarchy, and which slots are free for new data.
Most file systems rely on a handful of shared building blocks. A cluster or block is the smallest unit of storage the file system will hand out to a file, even if the file only needs part of it, which is why a 1 KB file can still take up 4 KB or more on disk. A table or index, such as the File Allocation Table in FAT systems, the Master File Table in NTFS, or inodes in Linux and Unix-style systems, records which clusters belong to which file and in what order. Many modern file systems also keep a journal, a running log of pending changes, so that if the power cuts out mid-write, the system can recover cleanly instead of leaving files corrupted.
FAT32: the oldest common denominator
FAT32 (File Allocation Table, 32-bit) grew out of the original FAT system Microsoft introduced in 1977, and became widely available with Windows 95 OSR2 in 1996, extending the older FAT16 format. It remains one of the most widely compatible file systems, readable by Windows, macOS, Linux, games consoles, cameras, drones, and most car stereos and TVs with a USB port.
Its defining limitation is a maximum individual file size of 4 GB, because file sizes are tracked in a 32-bit field. That means a single video file just over 4 GB, a common size for an hour of 4K footage, will not fit on a FAT32 drive. For decades, Microsoft’s own formatting tool in Windows also capped FAT32 volumes at 32 GB, even though the FAT32 specification itself supports volumes up to 2 TB with standard 512-byte sectors, or up to 16 TB with 4,096-byte sectors. In 2026, Microsoft began rolling out an update to Windows 11 that raises the command-line formatting limit from 32 GB to 2 TB, though this only changes the maximum volume size, not the 4 GB per-file ceiling, which is unchanged.
Typical modern uses of FAT32 include small USB flash drives, SD cards under about 32 GB, digital camera memory cards, and firmware update drives for routers and other embedded devices, largely because almost every device in the world can read it.
exFAT: FAT32’s successor for large files
Microsoft introduced exFAT (Extended File Allocation Table) in 2006 with Windows Embedded CE 6.0, aimed at flash memory such as SD cards and USB drives that had outgrown FAT32’s limits. exFAT keeps FAT32’s simplicity and broad compatibility but replaces the 32-bit size fields with 64-bit ones. According to Microsoft’s own published specification, exFAT uses 64 bits to describe file size, which removes the 4 GB per-file ceiling and supports very large storage devices through clusters as large as 32 MB.
In practice, exFAT’s theoretical file and volume size limits both reach into the exabyte range, though 512 TB is generally treated as the realistic upper bound for both. The SD Association has adopted exFAT as the mandatory file system for SDXC cards larger than 32 GB and for SDUC cards, so most SD cards sold today above 32 GB in capacity, including those used in phones, cameras, and drones, are formatted in exFAT out of the box. Microsoft kept exFAT proprietary and patent-protected for years, which slowed adoption on Linux; it published the specification openly in August 2019, after which royalty-free open-source support followed and mainline Linux kernel support for reading and writing exFAT arrived without needing separate driver installation.
exFAT does not include a journal, so unlike NTFS or ext4 it offers little built-in protection against corruption from sudden power loss, which matters for anyone using it on drives that get unplugged casually, a common situation with shared USB drives in offices, internet cafés, and schools.
NTFS: the default for modern Windows
NTFS (New Technology File System) has been Microsoft’s primary file system since Windows NT 3.1 in 1993, and it remains the default for internal drives on every current version of Windows. According to Microsoft’s official documentation, NTFS volumes on Windows Server 2019 and later, and Windows 10 version 1709 and later, can be as large as 8 petabytes; earlier versions of Windows support volumes up to 256 terabytes.
Maximum file size depends on the Windows version and cluster size. Microsoft’s Open Specifications documentation states that for Windows 8.1 and later, the maximum NTFS file size is calculated as (2^32 × cluster size) minus 64 KB, while earlier versions such as Windows 7 capped individual files at 16 TB. With the larger 2 MB cluster size supported since Windows 10 version 1709, that ceiling extends to roughly 8 PB, matching the volume limit.
NTFS supports features that FAT-based systems lack: file and folder permissions, disk quotas, transparent compression, encryption through the Encrypting File System, and integration with BitLocker drive encryption. It also uses a transaction log, so if a system crashes mid-write, NTFS can restore consistency automatically on the next startup. This makes NTFS well suited to internal system drives but a poor default choice for drives you regularly move between Windows, macOS, and Linux computers, since macOS can only read NTFS drives by default and cannot write to them without third-party software, and Linux support, while functional, is less consistent across distributions than support for FAT-based formats.
Microsoft’s newer Resilient File System, ReFS, is worth a brief mention: it is available on Windows Server and some Windows 10/11 Pro for Workstations editions, is designed to better resist data corruption through built-in checksums, and can in principle scale to a single file of up to 2^64 − 1 bytes, though it is not intended to replace NTFS on ordinary consumer PCs.
ext2, ext3, and ext4: the Linux mainstream
The extended file system family, ext, has been Linux’s default lineage for decades. ext4, the current version, is described in Red Hat’s enterprise documentation as a scalable extension of ext3, supporting a maximum individual file size of 16 terabytes on current Red Hat Enterprise Linux releases, with file systems able to grow to around 50 terabytes in practice, even though the format’s theoretical ceiling is closer to 1 exabyte. Documentation from Oracle Linux confirms the same practical figures: a 16 TiB maximum file size and roughly a 1 EiB theoretical file system size for ext4 under supported kernels.
ext4 introduced extents rather than the older block-mapping scheme, which improves performance with large files and reduces fragmentation, and it removed ext3’s 32,000 subdirectory limit, allowing effectively unlimited subdirectories per folder. Because ext4 is the default file system on most desktop and server Linux distributions, and Android historically used it internally for years before many devices moved to F2FS, ext4 is one of the file systems most people in the world interact with without realizing it, since a large share of budget and mid-range Android phones, especially popular across Africa and South and Southeast Asia, are built on the Linux kernel.
APFS: Apple’s current file system
Apple introduced the Apple File System (APFS) at its 2016 developer conference as the replacement for HFS+, rolling it out to 64-bit iOS devices in March 2017 with iOS 10.3, and to Mac computers in September 2017 with macOS High Sierra. Apple’s own developer documentation describes APFS as a 64-bit file system supporting more than 9 quintillion files on a single volume, and it is used across macOS, iPadOS, iOS, tvOS, watchOS, and visionOS, meaning the same underlying format now spans everything from the Apple Watch to the Mac Pro.
APFS was built specifically for flash and solid-state storage, and its notable features include native strong encryption, “clone” files that let the system duplicate a file instantly without doubling storage use until the copies diverge, space sharing between volumes in the same container, snapshots, and fast directory sizing. According to Apple’s technical documentation, the design supports file sizes up to 8 exabytes. Apple has said it plans to fully retire support for encrypted HFS+ volumes in a future macOS release, continuing the multi-year transition away from the older format.
HFS+ and the original Mac file systems
Before APFS, Apple used HFS+ (Hierarchical File System Plus), introduced in 1998 with Mac OS 8.1 as an upgrade to the original 1985 Hierarchical File System (HFS). HFS+ moved from 16-bit to 32-bit allocation block addressing, letting it manage up to 4,294,967,296 blocks instead of HFS’s 65,536, which sharply reduced wasted space on larger drives and allowed a theoretical file and volume size of up to 8 exabytes. Original HFS, by contrast, was limited to a 2 GB maximum file size and 65,535 files per volume, and Apple’s own knowledge base documented a 2 GB per-file limit and a 32,767-file limit within a single folder.
HFS+ is now largely a legacy format, still readable by macOS and used in some older external drives and backup images, but it has been superseded by APFS for all newer Apple devices and default formatting choices.
Btrfs and ZFS: modern journaling and pooled storage systems on Linux and Unix
Two more file systems come up often in servers, network-attached storage (NAS) boxes, and enthusiast home labs: Btrfs and ZFS.
ZFS was originally developed by Sun Microsystems for Solaris, released in 2005, and is now maintained under the OpenZFS project for Linux, FreeBSD, and other platforms after Oracle’s acquisition of Sun. It combines the roles of a file system and a volume manager, and offers built-in RAID-like redundancy, checksums on all data to detect silent corruption, compression, deduplication, and snapshots. Independent 128-bit addressing gives it an enormous theoretical ceiling, generally cited at 16 exbibytes for a single file. TrueNAS and similar NAS platforms are commonly built around ZFS specifically for that data-integrity feature set.
Btrfs (B-tree file system) is a copy-on-write Linux file system with similar goals to ZFS, including snapshots and built-in RAID options, and it is used as the default or a supported option on several NAS operating systems, including Synology’s DSM. Independent 2024 academic benchmarking covering ext4, XFS, Btrfs, and ZFS at scale found meaningful differences in real-world reliability and performance under very large file counts, so professional storage decisions increasingly weigh benchmark data alongside theoretical size limits rather than the limits alone.
XFS deserves a mention alongside these two: it is a mature, high-performance journaling file system originally developed by Silicon Graphics and now a default option on Red Hat Enterprise Linux and other enterprise Linux distributions, well regarded for handling very large files and high-throughput workloads, though it cannot be shrunk after creation, only grown.
Network file systems: how computers share storage over a network
The file systems described above manage data on a single physical device. A different category, network file systems, defines how a computer accesses files that physically live on another machine over a network. The two dominant protocols are NFS and SMB.
NFS (Network File System) was originally developed by Sun Microsystems and is the standard for sharing files between Unix and Linux machines. It has gone through several versions defined in Internet Engineering Task Force RFCs; NFS version 4.2, documented in RFC 7862, added features such as server-side copy, space reservation, and support for sparse files, building on the security, compound operations, and internationalization improvements introduced in NFS version 4. NFS is the default choice for shared storage between Linux servers and is common in cloud and enterprise environments, though it also appears in some consumer NAS devices for Linux and macOS clients.
SMB (Server Message Block), sometimes still referred to by its older name CIFS (Common Internet File System), is the protocol Windows uses to share files, and it also has broad support on macOS and Linux through the open-source Samba project. According to Microsoft’s own documentation, the protocol has moved through several dialects: SMB 1.0 in Windows 2000 through Server 2003; SMB 2.0, introduced with Windows Vista Service Pack 1 in 2008, which cut the number of subcommands from over 100 to fewer than 20 to reduce network chatter; SMB 2.1 with Windows 7; SMB 3.0 with Windows 8 and Server 2012, which added multichannel connections and end-to-end encryption; and SMB 3.1.1, the current dialect, released with Windows 10 and Server 2016, which added pre-authentication integrity checks and, more recently, support for AES-256 encryption starting with Windows 11 and Windows Server 2022. Microsoft has been actively phasing out the original SMB1 dialect for security reasons; it is not installed by default in current Windows and Windows Server releases, and Microsoft’s guidance recommends updating any remaining SMB1-only servers rather than re-enabling the old protocol.
For everyday users, the practical difference is simple: NFS shares are the norm when Linux or Unix machines talk to each other, while SMB is what lets a Windows laptop, a Mac, and a network-attached storage box all see the same shared folder on a home or office network.
Optical discs, flash-optimized, and other file systems worth knowing
A few other formats round out the picture, especially for anyone dealing with older media, flash storage, or specialized devices.
ISO 9660 and its extensions (Joliet, Rock Ridge) are the standard file systems for CD-ROMs, defining how files and folders are laid out on read-only optical discs. UDF (Universal Disk Format) succeeded ISO 9660 for DVDs and Blu-ray discs and for some rewritable optical and flash media, and it is also the format many USB flash drives use when formatted for cross-platform video playback on TVs and Blu-ray players.
F2FS (Flash-Friendly File System) was developed by Samsung and merged into the Linux kernel in version 3.8, released in February 2013, specifically to suit the way NAND flash memory behaves, since flash storage has no spinning platters, read/write heads, or the seek penalties that older file systems were designed around. F2FS uses a log-structured design and built-in wear-leveling awareness, and it became the default internal storage format on many Android phones, including devices from Samsung, Google, and other manufacturers, once Android moved away from ext4 for internal storage on supported models.
Older formats worth knowing about mostly for historical or compatibility reasons include FAT12, used on floppy disks and very small early volumes with a 12 MB maximum volume size, and FAT16, the predecessor to FAT32 used on early hard drives with a roughly 4 GB volume ceiling. Cloud storage systems such as Google Drive, Dropbox, and OneDrive are not, strictly speaking, file systems in the traditional sense; they are object storage or file-sync services built on top of file systems running in a data centre, though from a user’s point of view they behave much like a very large, always-online drive.
Comparing the essentials
| File system | Typical use today | Max file size (as documented) | Max volume size (as documented) |
|---|---|---|---|
| FAT32 | USB drives, older cameras, small SD cards, firmware drives | 4 GB | 2 TB (command-line, Windows 11, 2026); up to 16 TB with 4 KB sectors |
| exFAT | SDXC/SDUC cards, large USB drives, cross-platform external drives | Up to hundreds of TB in practice; spec supports far more | 128 PB (recommended up to 512 TB) |
| NTFS | Internal Windows system and data drives | Up to about 8 PB (Windows 10 v1709/Server 2019 and later) | Up to 8 PB (same versions) |
| ext4 | Linux desktops and servers, many Android internal storage setups | 16 TiB (practical, current enterprise Linux) | About 50 TiB practical; up to roughly 1 EiB theoretical |
| APFS | macOS, iOS, iPadOS, watchOS, tvOS, visionOS | Up to 8 EB | Limited mainly by container/storage size |
| HFS+ | Legacy Mac drives and backups | Up to 8 EB | Up to 8 EB |
| ZFS | NAS boxes, enterprise storage, data integrity-focused servers | 16 EiB | Enormous (128-bit addressing) |
| Btrfs | Linux, some NAS platforms (e.g., Synology) | 16 EiB | 16 EiB |
| NFS | Shared storage between Linux/Unix machines | Depends on underlying file system | Depends on underlying file system |
| SMB/CIFS | Shared storage across Windows, macOS, Linux (via Samba) | Depends on underlying file system | Depends on underlying file system |
Note that NFS and SMB are network protocols, not on-disk formats, so their practical size limits are usually set by whichever file system, such as NTFS, ext4, or ZFS, is actually storing the data on the server.
Myths and misconceptions, clearly marked
Several claims about file systems circulate widely online and deserve a direct correction.
Myth: FAT32 cannot handle volumes larger than 32 GB at all. Fact: the FAT32 specification itself supports volumes up to 2 TB with standard sectors; the 32 GB cap that many people encountered for years was a limitation Microsoft built into its own formatting tools, not the file system’s actual design, as confirmed by Microsoft’s 2026 update raising that command-line limit to 2 TB.
Myth: exFAT is unsafe or unreliable compared to FAT32. Fact: this is partly true but overstated; exFAT genuinely lacks a journal, so it is more vulnerable to corruption from sudden disconnection than a journaled system like NTFS or ext4, but this is a design trade-off for simplicity and flash-storage efficiency, not evidence of a broken or defective format, and it is the mandatory standard for SDXC/SDUC cards precisely because it works reliably within its intended use case.
Myth: converting a drive from FAT32 to NTFS or exFAT will preserve all your files automatically. Fact: while Windows offers a convert command that can change FAT32 to NTFS in place, exFAT conversions and cross-format changes generally require a full reformat, which erases the drive, so a backup beforehand is not optional caution but a practical necessity.
Myth: a file system “wears out” a flash drive faster if you use the “wrong” one. Fact: flash memory does have a limited number of write/erase cycles, but this is managed primarily by wear-leveling in the drive’s own controller or, in file systems like F2FS, by the file system’s design choices, rather than being something an ordinary user causes by picking FAT32 over exFAT for a USB stick.
Myth: NTFS drives can be freely written to on any Mac out of the box. Fact: macOS can read NTFS drives natively but cannot write to them without third-party software or workarounds, a limitation that still surprises many users moving drives between Windows and Mac computers.
Disclaimer and an invitation to correct us
File system specifications, especially maximum theoretical limits, are often described differently across official documentation, community wikis, and enthusiast sites, and practical limits change as operating systems and kernels are updated. Wherever this article cites a specific number, it reflects the most authoritative and recent source available at the time of research; some figures, particularly very large theoretical ceilings, are unlikely to ever be reached in real-world hardware and are included for completeness rather than practical planning. If you spot an error, an outdated figure, or a claim that needs a stronger source, please contact us so it can be corrected.
References
- Microsoft Learn, “NTFS overview” — https://learn.microsoft.com/en-us/windows-server/storage/file-server/ntfs-overview
- Microsoft Learn, “[MS-FSA]: Appendix A: Product Behavior” (NTFS and file system size limits by Windows version) — https://learn.microsoft.com/en-us/openspecs/windows_protocols/ms-fsa/4e3695bd-7574-4f24-a223-b4679c065b63
- Microsoft Learn, “exFAT File System Specification” — https://learn.microsoft.com/en-us/windows/win32/fileio/exfat-specification
- TechSpot, “Microsoft is removing 32GB size limit for FAT32 volumes, this time for real” — https://www.techspot.com/news/112058-microsoft-removing-32gb-size-limit-fat32-volumes-time.html
- PCWorld, “Windows 11 is finally removing the 32GB FAT32 capacity limit” — https://www.pcworld.com/article/3112871/windows-11-is-finally-removing-the-32gb-fat32-capacity-limit.html
- Hackaday, “Microsoft Finally Ups FAT32 Size Limit” — https://hackaday.com/2026/04/17/microsoft-finally-ups-fat32-size-limit/
- Red Hat Documentation, “Getting started with an ext4 file system,” Red Hat Enterprise Linux 9 — https://docs.redhat.com/en/documentation/red_hat_enterprise_linux/9/html/managing_file_systems/getting-started-with-an-ext4-file-system_managing-file-systems
- Oracle, “Maximum File and File System Size Requirements,” Oracle Linux 9 documentation — https://docs.oracle.com/en/operating-systems/oracle-linux/9/fsadmin/fsadmin-MaximumFileandFileSystemSizeRequirements.html
- Gentoo Wiki, “ext4” — https://wiki.gentoo.org/wiki/Ext4
- Apple Developer Documentation Archive, “Apple File System Guide: Introduction” — https://developer.apple.com/library/archive/documentation/FileManagement/Conceptual/APFS_Guide/Introduction/Introduction.html
- Apple Developer Documentation Archive, “Apple File System Guide: Frequently Asked Questions” — https://developer.apple.com/library/archive/documentation/FileManagement/Conceptual/APFS_Guide/FAQ/FAQ.html
- Apple Developer Documentation Archive, “Apple File System Guide: Features” — https://developer.apple.com/library/archive/documentation/FileManagement/Conceptual/APFS_Guide/Features/Features.html
- Steve’s Blog, “Hierarchical File System (HFS) Format Specifications,” citing Apple Knowledge Base Article 8647 — https://www.savagetaylor.com/2018/01/01/mac-os-standard-format-specifications-apple-kb-article-8647/
- IETF/RFC Editor, RFC 7862, “Network File System (NFS) Version 4 Minor Version 2 Protocol” — https://www.rfc-editor.org/rfc/rfc7862.html
- Microsoft Learn, “SMB features in Windows and Windows Server” — https://learn.microsoft.com/en-us/windows-server/storage/file-server/smb-feature-descriptions
- Microsoft Learn, “Detect, enable, and disable SMBv1, SMBv2, and SMBv3 in Windows” — https://learn.microsoft.com/en-us/windows-server/storage/file-server/troubleshoot/detect-enable-and-disable-smbv1-v2-v3
- Visuality Systems, “SMB Protocol Guide: Dialects, Ports, Security” — https://visualitynq.com/resources/articles/smb-protocol/
- TweakTown, “Samsung’s new file system, F2FS, is both flash-friendly and open source” — https://www.tweaktown.com/news/26118/samsung_s_new_file_system_f2fs_is_both_flash_friendly_and_open_source/index.html
- LinuxLinks, “F2FS – flash file system” — https://www.linuxlinks.com/f2fs-flash-file-system/
- Kanguru Solutions, “exFAT Limitations” — https://www.kanguru.com/community/knowledgebase/basic-usb-flash-drives-knowledge-base/exfat-limitations
