8 ms·

Siracusa will finally be happy. ;-) Sounds an awful like like ZFS (zero-cost clones, read-only snapshots) but could it be? I would imagine they'd start from sc

by cpr 10y ago

Siracusa will finally be happy. ;-) Sounds an awful like like ZFS (zero-cost clones, read-only snapshots) but could it be? I would imagine they'd start from scratch to due IP issues. Clearly this is immature technology they want to get out for testing/evaluation before it's fully adopted even into their own products. (See below.) - - - from the release notes - - - As a developer preview of this technology, there are currently several limitations: Startup Disk: APFS volumes cannot currently be used as a startup disk. Case Sensitivity: Filenames are currently case-sensitive only. Time Machine: Time Machine backups are not currently supported. FileVault: APFS volumes cannot currently be encrypted using FileVault. Fusion Drive: Fusion Drives cannot currently use APFS.

DannyBee 10y ago
"FileVault: APFS volumes cannot currently be encrypted using FileVault. " This one is confusing, because this is a logical volume feature. Not sure how or why APFS would ever care that some layer above it is encrypting stuff.
- Someone 10y ago
  On the one hand, that may be an artificial limitation. If this turns out to have some bug that overwrites the encryption keys, they could of course say "we warned you", but it would not be good PR. Also, if beta developers report intermittent smaller data-losing bugs, they might want to study the affected drives to see what went wrong with it. Not having encryption enabled on them will make that a tiny bit easier. On the other hand, if the new ability to partition drives with flexible partition sizes includes separate encryption keys per partition, and encryption/decryption is done by the block driver, they may have work to do to keep that block driver informed about what blocks should get encrypted with what key.
- leonroy 10y ago
  Could be for performance reasons rather than a technical blocker. That said AES-NI instructions do make encryption pretty fast so it's anyone's guess at this stage.
- tptacek 10y ago
  What would be really neat is if this is because FileVault is going to be filesystem-level when they ship APFS. It's wishful thinking, I know.
- CGamesPlay 10y ago
  Err, per [1] > APFS supports encryption natively. You can choose one of the following encryption models for each volume in a container: no encryption, single-key encryption, or multi-key encryption with per-file keys for file data and a separate key for sensitive metadata. APFS encryption uses AES-XTS or AES-CBC, depending on hardware. Multi-key encryption ensures the integrity of user data even when its physical security is compromised. [1] https://developer.apple.com/library/prerelease/content/documentation/FileManagement/Conceptual/APFS_Guide/GeneralCharacteristics/GeneralCharacteristics.html#//apple_ref/doc/uid/TP40016999-CH2-SW1 https://developer.apple.com/library/prerelease/content/docum...
  - DannyBee 10y ago
    I wonder why they went this route. Most apple volumes are already logical volumes (check diskutil list). FileVault itself is, right now, implemented as part of corestorage (see diskutil cs for the encryption/decryption commands). I assume they decided they just wanted to go the entire ZFS route and get rid of core storage in favor of a pool model, but still ...
TillE 10y ago
ZFS gobbles RAM and almost certainly couldn't be made to run acceptably on the Apple Watch. No, this seems like something developed from scratch to meet their particular needs.
- ryao 10y ago
  ZFS needs very little memory to run. Performance is definitely better with more RAM, but the overwhelming use of memory in ZFS is for cache. Eviction is not particularly efficient due to the cache being allocated from the SLAB allocator, but that is changing later this year. Getting ZFS to run on the Apple Watch is definitely possible. I am not sure what acceptably means here. It is an ambiguous term.
  - kstrauser 10y ago
    The FreeBSD wiki (https://wiki.freebsd.org/ZFSTuningGuide https://wiki.freebsd.org/ZFSTuningGuide) still claims: > To use ZFS, at least 1 GB of memory is recommended (for all architectures) but more is helpful as ZFS needs lots of memory. Is that inaccurate?
    - KaiserPro 10y ago
      because it has its own FS cache.
    - mioelnir 10y ago
      I've ran the initial FreeBSD patchsets for ZFS support on a dual-p3 with 1.5g ram, so 1g for a recent version should be more than doable. ZFS on FreeBSD has become a lot better with low-memory situations. There are two additional things to consider. ZFS uses RAM mostly for aggressive caching to cover over both spinning disks and the iops tradeoff vdevs make over traditional raid arrays. Thus low memory is not such a big deal if you have a pool with a single SSD or NVMe device. The other point to consider is that on at least any non-Solaris derived platforms, the VFS layer does not speak ARC. So data is copied from an ARC object into a VFS object, taking up space in both. If you are able to adopt your platform to use the ARC as direct VFS cache, you can save RAM that way as well.
legulere 10y ago
Another reason is that ZFS just isn't fitting for Apple devices. It's memory hungry and energy hungry and has several limitations compared to HFS+ like not being able to be resized.
- ryao 10y ago
  ZFS will be happy on a system with only 512MB of RAM, no matter how much storage it manages (assuming only 1 pool). It does need more RAM than UFS, but the amount is not notable unless we are talking about systems with 32MB of RAM. Being energy hungry relative to UFS and others is likely true due to things like checksum calculations and compression, but there is no way to implement these things without needing more cycles to compute them.
  - mschuster91 10y ago
    > Being energy hungry relative to UFS and others is likely true due to things like checksum calculations and compression, but there is no way to implement these things without needing more cycles to compute them. Not so true now - people have added encryption and compression instructions to CPUs. I'd be surprised if Apple couldn't ask Intel for a couple opcodes, and with the mobile platforms they do it anyway.
    - wstrange 10y ago
      But why couldn't ZFS also take advantage of those opcodes?
      - creshal 10y ago
        I'd be surprised if it doesn't. AVX is really good at speeding up compression/checksumming algorithms, and AESNI is standard in most AES implementations nowadays.
      - mschuster91 10y ago
        Because not every opcode is made public. The "usual suspects" for SIMD and encryption are public, yes, but nothing stops Intel from adding opcodes so highly specialized that they essentially represent the exact program code of the filesystem.
cmyr 10y ago
There is going to be some absolutely crazy sex in the Siracusa household this evening.
- bwilliams18 10y ago
  He's in SF w/ Casey and Marco... :/
  - lostlogin 10y ago
    Are they not his householders?
m_mueller 10y ago
> Siracusa will finally be happy. ATP will just be a concert of dings.

by cpr 10y ago

DannyBee 10y ago

"FileVault: APFS volumes cannot currently be encrypted using FileVault. " This one is confusing, because this is a logical volume feature. Not sure how or why APFS would ever care that some layer above it is encrypting stuff.

Someone 10y ago
On the one hand, that may be an artificial limitation. If this turns out to have some bug that overwrites the encryption keys, they could of course say "we warned you", but it would not be good PR. Also, if beta developers report intermittent smaller data-losing bugs, they might want to study the affected drives to see what went wrong with it. Not having encryption enabled on them will make that a tiny bit easier. On the other hand, if the new ability to partition drives with flexible partition sizes includes separate encryption keys per partition, and encryption/decryption is done by the block driver, they may have work to do to keep that block driver informed about what blocks should get encrypted with what key.
leonroy 10y ago
Could be for performance reasons rather than a technical blocker. That said AES-NI instructions do make encryption pretty fast so it's anyone's guess at this stage.
tptacek 10y ago
What would be really neat is if this is because FileVault is going to be filesystem-level when they ship APFS. It's wishful thinking, I know.
CGamesPlay 10y ago
Err, per [1] > APFS supports encryption natively. You can choose one of the following encryption models for each volume in a container: no encryption, single-key encryption, or multi-key encryption with per-file keys for file data and a separate key for sensitive metadata. APFS encryption uses AES-XTS or AES-CBC, depending on hardware. Multi-key encryption ensures the integrity of user data even when its physical security is compromised. [1] https://developer.apple.com/library/prerelease/content/documentation/FileManagement/Conceptual/APFS_Guide/GeneralCharacteristics/GeneralCharacteristics.html#//apple_ref/doc/uid/TP40016999-CH2-SW1 https://developer.apple.com/library/prerelease/content/docum...
- DannyBee 10y ago
  I wonder why they went this route. Most apple volumes are already logical volumes (check diskutil list). FileVault itself is, right now, implemented as part of corestorage (see diskutil cs for the encryption/decryption commands). I assume they decided they just wanted to go the entire ZFS route and get rid of core storage in favor of a pool model, but still ...

TillE 10y ago

ZFS gobbles RAM and almost certainly couldn't be made to run acceptably on the Apple Watch. No, this seems like something developed from scratch to meet their particular needs.

ryao 10y ago
ZFS needs very little memory to run. Performance is definitely better with more RAM, but the overwhelming use of memory in ZFS is for cache. Eviction is not particularly efficient due to the cache being allocated from the SLAB allocator, but that is changing later this year. Getting ZFS to run on the Apple Watch is definitely possible. I am not sure what acceptably means here. It is an ambiguous term.
- kstrauser 10y ago
  The FreeBSD wiki (https://wiki.freebsd.org/ZFSTuningGuide https://wiki.freebsd.org/ZFSTuningGuide) still claims: > To use ZFS, at least 1 GB of memory is recommended (for all architectures) but more is helpful as ZFS needs lots of memory. Is that inaccurate?
  - KaiserPro 10y ago
    because it has its own FS cache.
  - mioelnir 10y ago
    I've ran the initial FreeBSD patchsets for ZFS support on a dual-p3 with 1.5g ram, so 1g for a recent version should be more than doable. ZFS on FreeBSD has become a lot better with low-memory situations. There are two additional things to consider. ZFS uses RAM mostly for aggressive caching to cover over both spinning disks and the iops tradeoff vdevs make over traditional raid arrays. Thus low memory is not such a big deal if you have a pool with a single SSD or NVMe device. The other point to consider is that on at least any non-Solaris derived platforms, the VFS layer does not speak ARC. So data is copied from an ARC object into a VFS object, taking up space in both. If you are able to adopt your platform to use the ARC as direct VFS cache, you can save RAM that way as well.

legulere 10y ago

Another reason is that ZFS just isn't fitting for Apple devices. It's memory hungry and energy hungry and has several limitations compared to HFS+ like not being able to be resized.

ryao 10y ago
ZFS will be happy on a system with only 512MB of RAM, no matter how much storage it manages (assuming only 1 pool). It does need more RAM than UFS, but the amount is not notable unless we are talking about systems with 32MB of RAM. Being energy hungry relative to UFS and others is likely true due to things like checksum calculations and compression, but there is no way to implement these things without needing more cycles to compute them.
- mschuster91 10y ago
  > Being energy hungry relative to UFS and others is likely true due to things like checksum calculations and compression, but there is no way to implement these things without needing more cycles to compute them. Not so true now - people have added encryption and compression instructions to CPUs. I'd be surprised if Apple couldn't ask Intel for a couple opcodes, and with the mobile platforms they do it anyway.
  - wstrange 10y ago
    But why couldn't ZFS also take advantage of those opcodes?
    - creshal 10y ago
      I'd be surprised if it doesn't. AVX is really good at speeding up compression/checksumming algorithms, and AESNI is standard in most AES implementations nowadays.
    - mschuster91 10y ago
      Because not every opcode is made public. The "usual suspects" for SIMD and encryption are public, yes, but nothing stops Intel from adding opcodes so highly specialized that they essentially represent the exact program code of the filesystem.

cmyr 10y ago

There is going to be some absolutely crazy sex in the Siracusa household this evening.

bwilliams18 10y ago
He's in SF w/ Casey and Marco... :/
- lostlogin 10y ago
  Are they not his householders?

m_mueller 10y ago

> Siracusa will finally be happy. ATP will just be a concert of dings.