ZFS developer analyzes of the good and bad in Apple’s new APFS file system

“I’m not sure Apple absolutely had to replace HFS+, but likely they had passed an inflection point where continuing to maintain and evolve the 30+ year old software was more expensive than building something new,” Adam H. Leventhal writes for Ars Technica. “APFS is a product born of that assessment.”

“There are some seemingly absent or ancillary design goals: performance, openness, and data integrity. Squeezing the most IOPS or throughput out of a device probably isn’t critical on watchOS, and it’s relevant only to a small percentage of macOS users,” Leventhal writes. “It will be interesting to see how APFS performs once it ships.”

“It’s a shame that APFS lacks checksums for user data and doesn’t provide for data redundancy. Data integrity should be job one for a file system, and I believe that’s true for a watch or phone as much as it is for a server,” Leventhal writes. “APFS will be an improvement at stability for Apple users of all kinds, on every device. There are some clear wins and some missed opportunities. Now that APFS has been shared with the world, the development team is probably listening. While Apple is clearly years past the decision to build from scratch rather than adopting existing modern technology, there’s time to raise the priority of data integrity and openness.”

Tons more in the full article here.

MacDailyNews Take: Yes, there’s still time for improvement and refinement.

SEE ALSO:
APFS: What Apple’s new Apple File System means to you – June 24, 2016
APFS: New Apple File System promises more speed, flexibility, reliability – June 17, 2016
The feds’ll hate this: Apple’s new APFS file system ‘engineered with encryption as a primary feature’ – June 14, 2016
Buh-bye HFS+, hello APFS (Apple File System) for macOS! – June 14, 2016
Apple can do better than Sun’s ZFS – October 26, 2009
Apple discontinues ZFS project, turns attention to own next-gen file system – October 24, 2009
Apple’s Mac OS X Snow Leopard Server’s ZFS goes MIA – June 9, 2009

19 Comments

    1. I think the OP has a point insofar as data integrity checks are not front and centre in APFS.

      One of the biggest problems with HFS+ was its lack of features to detect and correct data integrity. Bitrot can go undetected in HFS+ volumes completely unbeknownst to the user, which is disturbing.

      The thrust of the article is that Apple has no explicit data integrity measures like checksums in APFS, which would seem odd given that they are an effective way to monitor data integrity with only a small additional footprint.

      The excuse of “our file system doesn’t have bitrot” doesn’t cut it; Apple are yet to provide an adequate explanation.

      1. I would guess that Apple has made design tradeoffs based on their clear understanding of how their devices are used.

        Preventing bit rot is a very useful feature if you’re building server infrastructure. Apple knows macs and iOS devices (as well as watchOS and tvOS devices) are not being used for that. Apple isn’t even building their own mass storage solutions based on Apple devices.

        1. Preventing bit rot is also important for one’s personal photos, too … and lest I be too cynical, Apple’s solution there is for us to simply pay them a mere $100/month per 1TB for iCloud, right?

          1. I agree. Data integrity is critical even for those that have just their long term personal photos to maintain. (Apple shipped its first digital camera — QuickTake — back in 1994. For individual who have images from that Apple camera for the past 22 years data integrity can be an issue.)

            However…
            No sane person (even the bean counters at Apple) expect you to back up your computers to the level of one or more TB to iCloud. You may have a “backup” of critical information there, but you won’t be doing full backups of multi TB drives. (Simple arithmetic shows that it will take over three hours per TB — under optimal circumstances with a 1 Gbps connection and 10x that for a 100 Mbps connection.)

            1. True, it does sound insane to contemplate paying for 1TB (or more) of iCloud backup space …

              … but in counterpoint, this is an incremental, “In-App” purchase temptation that starts when one caves on paying for anything more than Apple’s free 5GB allocation…and that’s where customers start their rationalization, at: “Oh, it is only $1 per Month”.

              With how Apple has restructured iPhoto into Photos, that consumption is now more than merely ‘critical’, but can include every photograph in that customer’s collection. Yup, all twenty years’ worth.

              Finally, it isn’t the upload that’s the bandwidth performance concern, since that will typically be spread out over months/years (and mostly in the background) … it is the (event X happened) data restoration download that will be the Bandwidth Pain Hammer…try calculating the restore time for Grandma who’s still got DSL.

    1. “Why is there an assumption that what his developer is whining about won’t be addressed?”

      Apple user since 1983, Mac since 1988

      As a matter of fact I DO have a right to question Apple. I make my living on Macs, do YES the details really do matter.

      Have a nice day

    1. As the guys behind it have a damn good track record while yours is totally unknown, I rather feel I know who I will be issuing the dumb award to in this contest. I certainly think you would have to be a fool to think they hadn’t considered the consequences of decisions they take in their work, when they would be more aware than you or I, of pre existing solutions to the issues stated. ZFS was well known by Apple and rejected for reasons we are not clear about, but had its own issues and criticisms so the author is by no means the fount of all knowledge on the subject. Others may feel differently and at this early stage there is much to learn. Let’s at least save the knee jerk reactions where they belong and go into the matter more deeply before jumping to conclusions.

  1. There are two suggestions at the start the report that strike me as being very wide of the mark with regard to Apple.

    “I’m not sure Apple absolutely had to replace HFS+ … ”
    Apple doesn’t delay doing things until it absolutely has to do them. Instead Apple changes things when they have a better solution. Indeed they often get a lot of criticism for changing things before most users recognise that change is beneficial.

    ” … but likely they had passed an inflection point where continuing to maintain and evolve the 30+ year old software was more expensive than building something new,”
    The implication there is that Apple chooses the cheapest option. The reality couldn’t be more different. Apple spends a huge amount of development money on things that most users would think were of no consequence and which most of their rivals would never bother with, but it’s that obsessive attention to detail and having every element of existing products subject to continual scrutiny and review that helps to set Apple apart from everybody else.

    1. Thanks, alanaudio! You post flags some flaws in the article that blow on past most people. Microsoft users, for example, would take those statements for granted as the standard way of doing business.

      I used to love hearing about new Apple technologies. But increasingly the initial responses that pop up are kneejerk reactions – it does not do this, it does not do that, etc. – despite the fact that it may just be a rumor concocted by some cyber journalist with some blank space to fill. Even when it is a pre-release version of a real product, the pundits pile on with their self-proclaimed expertise in all aspects of technology. It just takes the fun and excitement out of it.

      I am looking forward to APFS. I will not be an early adopter, though. I will give the early adopters at least three to six months to work out the major bugs before I make the transition. After all, this is my data and the file system is critical.

      Note for 2017:
      I don’t want to hear people whining about data loss after jumping onto a new file system without properly backing up their data. Back up thoroughly before transitioning to APFS and maintain those original backups until you are certain that your APFS volume is bit-for-bit complete! Anyone who comes to this forum whining about lost data will receive no sympathy.

    2. I don’t think he meant they chose the cheapest option, I think he meant they hit a point where it made more sense from a design standpoint and time investment to create a new one.

      Its a common problem in software development. You hit that point where you realize its less work and pain to just start over from scratch then to try and maintain the existing monster.

  2. I kind of assume this new file system is built with the Car in mind. Also smooth use of cloud data across multiple systems. Gotta look where the puck is going.

Reader Feedback

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