Building Your Own Fusion Drive

We missed the original announcement, but Apple unveiled more than just the iPad Mini at their last event. They’ve got a new storage system called Fusion Drive which is supposed to combine the access speeds of solid state with the storage density of platter drives. When you look just under the surface what you’re really seeing is a disc drive with grossly enlarged cache in the form of an SSD drive. How about moving from the 64 MB or so of cache seen on many large hard drives today to something like 64GB?

Well you don’t have to wait for Apple to do it. [Patrick Stein] gave it a shot using command line tools to combine an SSD with a physical drive. Sure, it’s not an all-in-one solution, but it is a pretty good proof. The linchpin that will really make it possible is a low-level driver that can handle the caching on the SDD, while ensuring that the data eventually makes it to the platter for long-term storage.

[via Engadget]

46 thoughts on “Building Your Own Fusion Drive

    1. While your response makes complete sense based on how hackaday describes FusionDrive, in reality, FusionDrive doesn’t act the same way as hybrid hard drives. FusionDrive stores the files (whole application bundles, documents, etc on a file level) either on the SSD, or on the hard drive. It has some logic built in that supposedly detects how often you use a file, and moves that file to either the hard drive or SSD based on frequency of use and throughput (it’s not going to store your mp3/mp4 files on the SSD since they’re 128kbps-20,000kbps and single threaded access requests, but it should choose to store applications or maybe mysql databases on the SSD. Cool technology, except it means that even though you’re using two drives, if you lose one, you lose all the data on it. I’m not sure if it’s possible with any existing tools to recover files from the surviving drive, but only the files stored on it would be recoverable.

    2. So what? It’s not the same: the data on a Fusion drive is more intelligently managed. It’s not merely using SSD as just a buffer. The SSD space is *added* to the HDD space (and not just “in front” of it) and the software decides if files should be kept on SSD (often accessed) or the HDD (less often accessed).

    1. Why would Apple lawyers care at all? He used Apple software on Apple hardware. Just because he’s doing something that’s not supported by Apple has nothing to do with any sort of infringement.

    1. Thank you! Lots of people seem to be confused about the difference. I’ve been using cache for a couple years now (ZFS) but for the typical consumer, this is actually some really cool stuff… assuming they’re doing backups.

      1. So why didn’t you just buy a Seagate Momentus XT Hybrid Drive?

        It’s not a cache either: “Testing confirms that the Momentus XT SSHD compares favorably to SSD—without sacrificing capacity”

        (the without sacrificing capacity means it’s not caching)

      1. And the MP3 player (Diamond RIO and many others before the original iPod), the smartphone (Handspring Visor with phone module and some other phones with touchscreens and minimal app collections before that came long before the iPhone), the PDA (Apple’s Newton was first but a failure, then Palm made it work by making it small enough then Apple “reinvented” it as the iPad Touch line – so it’s really the iPDA), supermini laptops (Toshiba’s entire Libretto line lived and died before the MacBook Air hit the scene, also the Palmax and a few others).

  1. It’s not a cache. It’s more like tiering in a server situation. If you have a 1TB drive and a 256GB SSD, you’re going to get a 1.25 TB Fusion Drive with pieces of files on either drive, only on both while a file is being moved and removed from the first once it has been confirmed to have been moved. Once you create this joined drive through the command line, it looks like Core Storage automatically determines where files go if it can determine that one of the drives is an SSD, or possibly just if it’s faster. Also, Apple talks about files being moved, but it looks like it’s moving blocks back and forth (John Siracusa talks about the difference in the link below).

    More information:
    http://arstechnica.com/apple/2012/11/fusion-drive-quick-look-our-predictions-confirmed/

    http://arstechnica.com/apple/2012/10/more-on-fusion-drive-how-it-works-and-how-to-roll-your-own/

    John Siracusa also talked about it in the last Hypercritical podcast: http://5by5.tv/hypercritical/92

  2. I saw them announced Big by SEAGATE and other Companies about 2 or 3 years ago…

    Apple “invented” the hybrid-harddrive – cool… a great idea from the company that brought us MP3 Players, Touchscreens, Mobile Phones and the Color White!

  3. Yah, it’s a tiered storage solution. It’s very easy to set up. I did it in an afternoon. I haven’t had time to do much testing yet, but I’m impressed with it so far.

    It really isn’t that similar to a hybrid drive, because everything happens in software within the OS it’s possible for more intelligent decisions to be made about the placement of blocks onto either device.

    1. My last few mainboards have had this “new” technology in the bios…nothing special. Never tried it…if I need speed on some things and storage on other things I just use two drives, but I guess this makes it easier for the kind of people who buy Apple products.

    1. But its quite similar. Granted, Intels technology uses the SSD as a cache instead of tiering it, but who really misses the SSD capacity when compared to the entire HDD? And with the unreliabilities of SSDs (several friends and me had SSDs die in the last months, across all vendors) its maybe even better to just use the SSD as cache, with more conservative settings you can just turn the SSD off and continue to use the disk alone.

    2. it isnt exactly cache…cache is temporary memory read from a disk…srt is written to the ssd so no disk reading is needed for the most common tasks, yet you still have disk copies of everything on the ssd. This is a huge plus if you are using encryption, as a failure of the ssd with the apple method would leave both drives unreadable. as for the extra space, the people who use this don’t have enough files to care about an extra few hundred gb.

  4. How about bringing back the sticky bit? The linux kernel has never supported it for individual files, but maybe now is the time.

    Maybe something like in ext5, we keep a copy of the partition’s superblock, the journal, and any file that has the sticky bit set in an associated SSD partition. The kernel can also keep an eye on the entire file system, and utilize the remainder of the SSD’s available space to automatically cache needed files for faster access.

  5. lenovo and many other companies have already done similar to this so whoop de doo, I’m seeing some laptops with 32GB m-sata drives for cheap but fast os startup and application loading, my younger brother has one of them although I prefer full ssd benefits :D

  6. cache algorithms on a SSD.. buy one so you can buy another within a year when wear-leveling gives up..

    It’s scary how little most consumers and vendors know about flash while being considered experts..

  7. Anyone who thinks this is similar to a hybrid drive hasn’t read up on it. First, there’s the intelligent management which, while not brand new, is new to consumers. The ones others are pointing to are not even in the same class as what Apple is selling. Second, the SSD portion in this case is 128gb, no other similar consumer level product has this much solid state storage.

    I’m no Apple fanboy, but in this case Apple has brought technology that was previously only available on enterprise level SANs (aka tiered storage) and brought it to consumers. This time Apple did actually do something new for consumers.

    1. Too bad they don’t know about wearing-leveling and this is going to discredit them down the road when it’s revealed these have short life-cycles..

      Flash is like oil, nobody knows how it works they just want it..

      1. Yes, I’m sure none of the engineers at Apple have heard of SSD wear. I’m still not sure SSD wear is anything but a theoretical concern. I’ve never had any of mine wear out and HDs are already, essentially, disposable. I have a drawer full of drives from 500gb to 1.5tb that I’m not using because they’ve been replaced. Who cares if an SSD won’t last 10 years? Whos using a 10 year old drive anyway?

      2. I’m also sure that the engineers at

        Seagate/Maxtor, Western Digital, OCZ, Microsoft

        are all so behind the times to this one company that doesn’t even specialize in hardware.

        Again, as someone above posted. Seagate Momentus XT Hybrid Drive: “SSHDs utilize Seagate flash management technology to deliver SSD-like performance. Testing confirms that the Momentus XT SSHD compares favorably to SSD—without sacrificing capacity”.

        Certainly sounds like they’re doing exactly the same thing — keeping a file on one or the other, so it’s not a cache.

    2. If you’re serious about having 128GB as the selling point, you’re fairly disillusioned. What happened when the first SSD hybrids — I’ll bet that they started with 16GB. Then someone else OMGBROKENSOUNDBARRIER 32GB… then WARGARBLLELELE 64GB… You might as well call it revolutionary or entirely new when processors go from 2.4GHz to 2.8GHz. To me, I call it the march of technology. Wake me up when SSDs hit 1TB for $200.

      Just a minor tweak to caching (yes, that’s all this is: caching. You never know which files will require to be on a faster drive). Ultimately, does it matter? Is there an appreciable speed gain? If not, what’s the point?

      You’re right though, a lot of APL fans throw out perfectly good hardware when it can be reused. I still have a couple of desktops lying around that are 8 years old that are good enough for what they do (simple web browsing, email, and the occasional video/movie).

      I hate seeing people being lied to. Just remember, you’re talking about a company that has specifically said to the public/courts that they’re foolish for believing that their ads and PR are truthful.

  8. RAID 0 + 16GB RAM Drive for cache = faster than any software based storage solution.

    Put your files you use constantly on the RAM drive, and near instant load times are available to you.

    Let the sheep purchase this new fangled technology and let them fool themselves into thinking its the greatest thing since sliced bread. Anyone with two neurons to rub together won’t be fooled as easily with the hype.

Leave a Reply to v8mediaCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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