Contribute
Register

Guide - Fusion Drive using tonymacx86 Tools & Chimera

Just a little side note: I am pleased that people are using the guide with success. I created the guide because I do that sort of thing. I however much prefer a 4 hard disk RAID 0 set. I have had great success with Seagate 500 GB 7200 RPM drives and I am running a 4 Hitachi 500 GB 2.5" 7200 RPM drives in a Raid 0 on my White Prodigy hack. I see nearly SATA SSD III speeds from my RAID 0 4 drive sets, they are bootable and I have 2 TB (less turn around room) to play with at far less cost then SSDs.

But regardless of what you put on line, you do need a data backup and recovery plan that works.

The idea is to have fun and learn. And OS X hacking this is my hobby.

neil
 
Just a little side note: I am pleased that people are using the guide with success. I created the guide because I do that sort of thing. I however much prefer a 4 hard disk RAID 0 set. I have had great success with Seagate 500 GB 7200 RPM drives and I am running a 4 Hitachi 500 GB 2.5" 7200 RPM drives in a Raid 0 on my White Prodigy hack. I see nearly SATA SSD III speeds from my RAID 0 4 drive sets, they are bootable and I have 2 TB (less turn around room) to play with at far less cost then SSDs.

But regardless of what you put on line, you do need a data backup and recovery plan that works.

The idea is to have fun and learn. And OS X hacking this is my hobby.

neil

At the cost of lots of power, space, and environment/resources. Plus you have 4 chances of failure instead of 2. Also I just calculated out the cost difference between the buying 4 of the cheapest 7200RPM SATAIII 500GB HDD and it came out to $239.96 These were Toshiba drives.

Then I priced out the absolute cheapest 2TB hdd it was a seagate barracuda green drive with a Kingston HyperX 120GB SSD SATA III drive. Those prices came to $199.98.

Granted a real RAID or even software RAID setup has its place but for 95% of users the Fusion drive is clearly the winner.
 
Good job on the clear outline Neil, but there's a fundamental issue which worries me.​
Has anyone actually shown that Fusion Drive works? Not that you can create a Fusion drive, we know that. But does it reliably do what everyone thinks it does?

When you put more data on there than will fit on the SSD you'll see it slow down as it starts using the HDD. But then when you repeatedly access data which is on the HDD has anyone shown that it gets migrated to the SSD?
I have seen a few tests so far that show that it doesn't and lots that don't show anything useful.
Maybe it works but what's the key to get the migration happening? Haven't seen it yet.

Has anyone been able to recover data from a crashed system? Repair the filesystem? I've seen some simple tests that produces fatal errors (as in blow all your data away and start again). This does not induce comfort.

For now I'll stick with separate SSD and HDD disks and put the data on each that needs to be on each.
I'm happy for you guys to experiment and be on the bleeding edge (presumably on systems you're not depending on for work) but be prepared for some pain. I'm just sayin'...

<Wanders off muttering about barge poles...>

I haven't configured it yet as my boot volume so I can't comment too much on day to day usage just yet. However I have setup a mock failure scenario and was able to recover the data that was on the still working volume using Data Rescue. So presumably if its a File system issue that diskwarrior will not be able to repair Data rescue should still get data from both disks.
 
Yes, it works. It's automatic and it's in the background. Anyone can use iostat to see what is happening on each drive. Where have you seen it reported that it doesn't work?
Some of the pages you linked to did have more-comprehensive testing than most of the tests I've seen, so that's good.

But maybe it's partly my brain not being comfortable that I can understand yet what it's doing under the covers (BTW, before I became a photographer I had a background in Unix internals including filesystems [incidentally, I just realised this week that my name is still in some of the ML documentation even though I never worked for Apple! :)] so I understand that my expectations are different than many users - I'm not used to the OS being a "black box" to me). It will be interesting to see how it pans out as more information comes to light as to the internals. Certainly Lloyd Chambers had some serious problems with Fusion Drive (a series of articles there).
As Neil mentions, whatever you do make sure you have backups and a recovery process. Of course, that should always apply even if you're not using Fusion Drive.

Any why are people trying to get Fusion Drive confused with Fear, Uncertainty, and Doubt (FUD is one of the primitive marketing methods used by many large companies - it's often associated with Microsoft). That just makes me laugh. :lol:
 
My Congratulations to Neilhart for an excellent guide.

I have loaded Fusion on my real MacMin1 2011 Server version running 10.8.2 non server version. The installed drives are a OCZ -Vertex3 and an Seagate ST750LX003 and it is faster in Fusion mode than the original OS loaded on the SSD and the Data on the Seagate. I am very impressed.

I have tried similar on the hack (using this excellent guide) and I do not see the performance gains I see on the real MacMini. I gave it two days and decided to revert.

The only difference is I am using Clover and total UEFI booting, which should be far better than anything bios and chameleon based. So forget the FUD and just enjoy what works best for you.
 
I followed this guide exactly but am having trouble booting up the fusion drive being stuck at "Still waiting for root device"

I have a Z68MA gigabyte UEFI bios board does that make a difference? The cloned partition boots up without any problems at all but just the fusion drive partition cannot boot up.
 
Good job on the clear outline Neil, but there's a fundamental issue which worries me.​
Has anyone actually shown that Fusion Drive works? Not that you can create a Fusion drive, we know that. But does it reliably do what everyone thinks it does?

When you put more data on there than will fit on the SSD you'll see it slow down as it starts using the HDD. But then when you repeatedly access data which is on the HDD has anyone shown that it gets migrated to the SSD?
I have seen a few tests so far that show that it doesn't and lots that don't show anything useful.
Maybe it works but what's the key to get the migration happening? Haven't seen it yet.

Has anyone been able to recover data from a crashed system? Repair the filesystem? I've seen some simple tests that produces fatal errors (as in blow all your data away and start again). This does not induce comfort.

For now I'll stick with separate SSD and HDD disks and put the data on each that needs to be on each.
I'm happy for you guys to experiment and be on the bleeding edge (presumably on systems you're not depending on for work) but be prepared for some pain. I'm just sayin'...

<Wanders off muttering about barge poles...>

So the fundamental question is sitll - is it worth it? The video'd speed tests on MrEngle's 2 Nov post show that the speed really isn't that great compared to Read/Write speeds of a straight SSD.

Is a Fusion Drive worth it?
 
So the fundamental question is sitll - is it worth it? The video'd speed tests on MrEngle's 2 Nov post show that the speed really isn't that great compared to Read/Write speeds of a straight SSD.

Is a Fusion Drive worth it?

Here is the my view, assuming i understand correctly. Some small overheard aside, writing and reading from a fusion drive should be very similar to doing the same from a straight SSD (in a SSD+HDD setup) until you fill up the SSD. The OS will always copy to the SSD part of the FD until it fills up then switch over to the HDD part. When that happens, of course you will suffer degraded read/writes. On a going forward basis after that point, the FD reserves at least 4GB of "cache" space on the SDD for quick writing. In the background, the OS should begin to learn (theoretically) which files you use more and elevate them to the SSD while demoting less used files to the HDD. If you were to read a file on a FD stored on the SSD part, i believe it should be almost as quick as reading from a straight SSD setup with the same file. As you write to the FD over time, it should be writing to the SSD part first and thus you should recognize similar benefits as if writing to a straight SSD.

When you ask if a FD drive is worth it, that is subjective. If you are not technical enough and/or do not want to manage a separate SSD and HDD setup where you could be storing a home directory on the HDD or symlinking between the SSD and HDD, then again theoretically a FD setup could be ideal. The OS will determine which files you are using more often and put them on the faster drive for you.

Not going into the risks of the FD setup but you should read about those too.

As to the "theoretical" part about FD, I have seen Ars test FD and it clearly seems like it works - but this is on a mac mini with FD pre installed. Other tests seem to show a BYO FD setup works as well although Lloyd at Mac Performance Guide questions whether it works. He also questions the methodology used to test i/o and speed results.

Going to try and test on my own to see whether or not I can get files to move based on usage.
 
Neil,

Great great guide. Followed it to a T and it worked swimmingly.

My only question is this. The Extra folder that I copied to my two helper partitions is not the same Extra folder that now resides on the Fusion volume - which was simply a clone of the original boot environment. If I want to make changes to the Extra folder or to the plist file, which one am i editing? And should i be keeping all three in sync? So any change to the hackintosh setup has to be pushed out to the hidden helper partitions?

Thanks again for the amazing guide.
 
Back
Top