skip to content

When a Mac's startup disk moved from HFS+ to APFS, which concrete behaviours changed for applications and for people administering the machine?

level: middleimportance: should knowfreq 45%

answer

  1. fixed partitions became a shared pool
  2. copies and rollback points got free
  3. journal replaced by copy-on-write
  4. one second became one nanosecond
  5. one HFS+ trick was dropped

basics

~20 s

APFS replaced HFS+ journaling with copy-on-write metadata, added clones, snapshots and volumes that share a container's free space, gave timestamps nanosecond resolution instead of one second, made encryption native and per-volume, and dropped support for directory hard links.

solid answer

~50 s

The visible changes cluster in five places. **Space**: HFS+ filesystems lived in fixed-size partitions; APFS volumes share a container's free pool and are never sized. **Cheap copies**: clones and snapshots simply did not exist on HFS+, so instant duplication and rollback points are new. **Crash consistency**: HFS+ journaled its metadata; APFS never overwrites in place and commits copy-on-write metadata at checkpoints instead. **Timestamps**: HFS+ stored one-second resolution, APFS stores nanoseconds — which quietly fixed a class of build tools that compare mtimes, and broke a few that assumed whole seconds. **Encryption**: FileVault on HFS+ was a separate CoreStorage layer; on APFS encryption is a native per-volume property with multiple keys. One removal matters too: APFS does not support directory hard links, the HFS+ trick that old Time Machine backups relied on. Case sensitivity is *not* a difference — both offer case-insensitive by default and case-sensitive as a format option.

go deeper

for a junior

Be able to name a few concrete differences — shared space instead of fixed partitions, instant duplication, snapshots — rather than saying APFS is simply newer or faster.

for a middle

Explain the mechanisms behind each difference: container plus volumes, copy-on-write metadata instead of journaling, nanosecond timestamps, native per-volume encryption in place of the CoreStorage layer.

for a senior

Talk about what actually bit people during the transition — timestamp-driven build and sync tools, capacity tooling confused by clones and snapshots, and backup formats that depended on directory hard links.

for a principal

Position it as a platform migration: Apple converted disks in place on hundreds of millions of machines, which constrains how much the on-disk semantics may change and explains the conservative choices, such as not checksumming user data.

## Why the change happened HFS+ dates from 1998 and inherited structures from HFS, designed for spinning disks and single-user Macs. Apple introduced APFS in macOS 10.12 Sierra, converted SSD startup disks automatically in 10.13 High Sierra, and extended conversion to Fusion and rotational drives in 10.14 Mojave. It was a from-scratch design aimed at flash storage, strong encryption and cheap metadata operations. Interviewers ask about the delta because it is a compact way to check whether you understand copy-on-write filesystems at all. ## Space model Under HFS+, one filesystem occupied one partition whose size was chosen at creation. Growing it meant shrinking a neighbour and moving data. APFS introduces the **container**, which owns the partition's blocks, and **volumes** inside it that allocate from a shared pool on demand. Nothing is sized in advance; free space is a container-wide property that every volume reports as its own. Administratively this is the biggest day-to-day difference: adding a volume is instant and non-destructive, and "resize the partition" stops being a task. ## Cheap copies: clones and snapshots HFS+ had neither. APFS adds: - **Clones** — a second file that shares the original's extents until either is written, so duplicating a huge file is a metadata operation. - **Snapshots** — read-only point-in-time images of an entire volume, created instantly, that pin the blocks reachable at that moment. Snapshots are what make macOS system updates reversible and Time Machine's local backups possible, and they are the reason free space on APFS no longer returns the instant you delete a file. ## Crash consistency HFS+ used a **journal**: intended metadata changes were written to a log first, then applied to the structures in place, and the log was replayed after an unclean shutdown. APFS takes the copy-on-write route instead — it never overwrites live metadata, writing new versions elsewhere and switching over atomically at a checkpoint. Either way you should not see a structurally inconsistent filesystem after a power cut, but the mechanism, and the write pattern it produces on flash, differ completely. APFS also adds checksums for its **metadata**. It is worth being precise here, because candidates overclaim: APFS does not checksum user file data the way ZFS or Btrfs do. ## Timestamps HFS+ stored timestamps at one-second resolution. APFS stores them with nanosecond resolution. This is the change most likely to reach ordinary application code. Build systems, rsync-style tools and caches that decide "is this newer than that?" by comparing mtimes get dramatically better discrimination — and code that assumed whole-second values, or that round-trips timestamps through a one-second representation, can misbehave. If you have ever seen a build system rebuild everything, or nothing, only on certain volumes, this is a candidate cause. ## Encryption On HFS+, FileVault 2 was implemented by **CoreStorage**, a logical volume layer stacked underneath the filesystem, which encrypted the volume as a block device. APFS makes encryption a native property of a volume, with a key hierarchy that supports a single volume key or per-file keys plus a separate metadata key. Practical consequences: encryption is per volume rather than per disk, it composes with the container model, and turning it on does not mean inserting a whole extra storage layer. ## The removal: directory hard links HFS+ supported hard links to *directories* — an unusual capability that Apple used to make early Time Machine backups compact, letting an unchanged folder in a new backup point at the previous one. APFS does not support them, which is one reason Time Machine's on-disk format changed and why backup destinations moved to APFS with snapshot-based mechanics in macOS 11. Regular file hard links and symlinks work on both. ## What did *not* change Candidates frequently list case sensitivity as an APFS innovation. It is not. HFS+ shipped in case-insensitive (default) and case-sensitive variants, and APFS does the same: the startup volume is case-insensitive but case-preserving, and you may format a volume case-sensitive if you need it. Likewise, extended attributes, resource-fork-style metadata and POSIX permissions carried over. ## Answering compactly If you get one minute: *space sharing instead of fixed partitions; clones and snapshots where there were none; copy-on-write metadata instead of a journal; nanosecond instead of one-second timestamps; native per-volume encryption instead of a CoreStorage layer; and directory hard links gone.*

  • Is case sensitivity one of the differences between HFS+ and APFS?
    No — that is a common false answer. Both filesystems ship in case-insensitive (default, case-preserving) and case-sensitive variants, and the macOS startup volume is case-insensitive in either era. What APFS makes easier is *getting* a case-sensitive filesystem: because volumes share a container, you can add a case-sensitive volume for a source tree in seconds without repartitioning or reformatting the disk.
  • Does APFS checksum file data the way ZFS does?
    No. APFS checksums its own metadata, so structural corruption is detectable, but user file data is not checksummed — Apple's reasoning being that the flash devices it targets do their own error correction. Claiming full end-to-end data integrity is a common overstatement; if the interview is about detecting silent data corruption, APFS is not the answer, and neither was HFS+.
  • Why did the loss of directory hard links matter?
    Early Time Machine relied on them: each new backup could hard-link an unchanged directory to the previous backup's copy, making incremental backups compact on HFS+. APFS does not support directory hard links, so that trick was unavailable and Time Machine's on-disk format changed, eventually moving to APFS destinations that use snapshots to get the same space sharing through copy-on-write instead.

saying these in an interview costs you the question

  • Says APFS introduced case sensitivity to the Mac
  • Claims APFS checksums user data end to end like ZFS
  • Describes APFS as journaled like HFS+
  • Thinks APFS volumes still have a fixed size like partitions
  • Says FileVault works the same way on both, just faster

context

resolving links…