WWDC 2014’s biggest disappointment

“Can you think of another technology from 1985 that is still central to your Mac? Me neither,” Robin Harris writes for ZDNet. “But HFS, the Mac/iOS file system, dates from 1985 – the early years of the Mac – and was lightly updated to become HFS+ in 1998, three years before Mac OS X debuted. HFS+ gave us 32 bit block addresses, longer file names and Unicode, but the underlying architecture remains 1985.”

“That’s why the biggest disappointment of this year’s WWDC is that no new file system was announced,” Harris writes. “Apple knows they have a problem: they announced ZFS on Mac Server back in 2007 before licensing issues and a lawsuit caused them to decommit.”

“The problem is that with users commonly storing millions of files, the bit rot inherent in storage – remember, the Universe hates your data – goes uncorrected and undetected,” Harris writes. “Until you try to access the file and you can’t.”

Read more in the full article here.


    1. Ok, we all agree that this is a weird way to look at the wonderful stuff that was introduced at WWDC. But once you get over the unfortunate approach that the author chose to bring up the subject, the issue of the core file system should not be dismissed. HFS+ is creaky and there are mature alternatives, notably ZFS. If you do a bit of research, you learn that one of the cool features of ZFS is the ability to treat many different physical volumes as a single logical volume. In an age when our files are scattered on multiple devices, and maybe even cloud systems, ZFS could conceivable unify our data. Imagine a Spotlight search – or even better, a Siri search – of all the places where our data resides. So, just saying, the question of whether Apple should upgrade its file system isn’t lame.

  1. After reading an article on file systems, I understand the complexity and conflicting needs for speed and parity check/checksums and the fact that file systems take a decade to fully develop.

    Apple is not going to upgrade its file system until it is dead certain if can go mainstream.

    That means a lot of time in Beta to have all the utilities and other programs updated as may be needed so we can properly work with the new file system.

    1. What makes you consider HFS to be mainstream?

      Apple introduced the Hierarchical File System in 1985. Despite some tweaks since, HFS+ was the last major improvement, implemented with Mac OS 8.1. It is a travesty that Apple has not made important updates since.

      Limitations include:
      – optimized for file listing, not for file accessability
      – although the Mac can support an 8 eB max volume size, non-Mac users are limited to 2 tB
      – 32-bit addressing
      – ineffcient handling of small file sizes
      – inability for more than one application to write at a time
      – Catalog List does not protect data corruption from affecting the entire system
      – non-Apple machines cannot write to an HFS+ drive
      – no checksum (ECC)
      – no automated data deduplication
      – OS X now implements stupid Windows-like filename extensions to associate file types

      Apple should do better.

      1. I agree with most of what you said, except for the filename extensions. OS X does file typing and app association much better than OS 9 did. More so, file extensions are needed now in a different era where files are stored and traded with systems that have no other means to associate file types. If you want to live in just a Mac bubble (don’t we all;) then you can simply hide extensions and not worry about them. But for those of us who need to work across many platforms, the extensions are an unfortunate requirement.

        They also have nothing to do with HFS+

      2. Also:
        “non-Apple machines cannot write to an HFS+ drive”

        No filesystem Apple develops is going to be writeable on non-Apple machines unless Apple open sources or licenses it, and then 3rd parties would need to implement it. There’s nothing stopping this from happening with HFS+ today from a technical perspective.

        That said, I’d love to see Apple develop a kick-ass file system and then evangelize the heck out of it so that it becomes the file system of choice for flash drives and other devices. It’s worth noting that as bad as HFS+ is, there are much worse things out there being used.

      3. He considers HFS+ to be mainstream because it is mainstream by definition.

        Mainstream: The ideas, attitudes, or activities that are regarded as normal or conventional (OS X Dictionary)

        As BoC indicated, Apple will not release a new filesystem until it is rock solid in every respect – i.e., ready to go mainstream

  2. The key considerations – is a new file system able to deliver a new and more fulfilling experience to Apple users ? Will this mean something insanely great that Apple is truly proud of doing… ? Will this help to gel our developers closer and have them motivated to expand the ecosystem?…

    I am afraid the writings are on the wall.

  3. Thank you, thank you, Robin Harris. So that’s why I’ve been so depressed for a few weeks. I can now force myself to get out of bed and attempt to make something out of my life again.

  4. I don’t get why WWDC 2014 is a flop because a file system that has been around for almost 30 years wasn’t changed this year just like the previous 28 years.

    I am outraged.

    1. That’s actually part of the problem. You can end up with a corrupt file being backed up. Unless you’re going to verify each individual file, the only way around this would be to endlessly archive, and then if you found a corrupt file, you’d need to search back in your archives to find the point in time after the file existed, but before it became corrupt.

      A more modern file system would resolve this issue, as well as speed things up, save space, and offer other functionality. As the author mentions, Apple realized this as least as far back as 7 years ago.

  5. Oracle owns ZFS now don’t they……Oracle and Apple have a history with Jobs, one would think if Apple was interested they would try and buy from Oracle.

  6. At least occasionally someone who knows what he is talking about exposes the truth about the once great company. Tim Cook is not doing anything other than keeping the hype going and going with showmanship, fanfare, and much ado about not much.

  7. I’m not going to make the case that I wouldn’t want Apple to come out with a new OS with better data integrity capabilities. However, to suggest that Apple has done nothing with HFS since 1985 is just silly.
    Apple has evolved HFS considerably over the years. Much of Apple’s design decisions have been in the name of backward compatibility. Further, many new features that are associated with the operating system are being handled through Core Storage.
    Since HFS+ in 1998, Apple has added Journaling (OS X 10.3), case sensitivity, ACL based file security, directory hard linking (10.5), file compression (10.6), logical volume encryption – through core storage (10.7).
    The direction Apple seems to be headed is in favor of core storage which sits on top of and is independent of the file system. As more APIs rely on core storage, swapping out file systems will eventually become trivial.

    1. I could easily list things I would like to have and indeed have done so in previous discussions. Tim Cooks promises of what we could expect was not specific, but did say we would get them throughout 2014. We are now approaching the end of the half year and still nothing tangible has been forthcoming. My biggest wish is for an larger iPhone that Is as good as the competion in both hardware and OS, not just a stretched 5s. WWDC did give me a little hope with features I have wished for on an iPhone for a long while, but I want something that goes beyond that, innovation not catching up with something I can have already.

Reader Feedback

This site uses Akismet to reduce spam. Learn how your comment data is processed.