HuntsMan75

Q: Replace Hard Drive with SSD

System Info: MacBook Pro running OS X 10.9, mid 2009, 13", PN MB991LL/A (2.53 GHz), 8GB RAM

 

I started having some problems with my system, mainly slow downs, especially when using Safari. I got a copy of Scannerz (http://scsc-online.com/Scannerz.html) and it confirmed drive platter damage. I know some of these units are supposed to have cable problems but Scannerz tested that too and found no problems with the cable, so this is a bonafide drive problem (it's 6 years old, you know).

 

I'd like to go ahead and replace it with an SSD and have some questions:

 

  1. Maybe I'm getting models mixed up but I seem to remember something about a drive thermal sensor on Apples not being compatible with generic drives. I know I read that somewhere I just don't remember where?
  2. I assume, if #1 above isn't a problem, that I can basically use any SATA SSD. Now I know I should get the fastest interface possible, but are there any caveats or incompatibilities that some brands of SSDs may present?
  3. What about stuff like trim and wear leveling. WIll this be in a driver for the SSD or does the OS take care of it automatically?
  4. Are there any brands/models to stay away from?
  5. Should I replace the cable too? I've heard they can get bad.

 

Thanks in advance.

MacBook Pro, OS X Mavericks (10.9.5)

Posted on Apr 4, 2015 12:06 PM

Close

Q: Replace Hard Drive with SSD

  • All replies
  • Helpful answers

Previous Page 2 of 15 last Next
  • by spudnuty,

    spudnuty spudnuty Apr 8, 2015 10:04 PM in response to spudnuty
    Level 5 (7,097 points)
    Mac OS X
    Apr 8, 2015 10:04 PM in response to spudnuty

    OK I'm answering myself because I want to address everyone.

    So addressing failure rates:

    There is a sudden failure due to power interruption for SSDs. Apparently only the Intel doesn't have this problem. However the operational advantages of an SSD are such a game changer that you gotta go with them. That's why Apple used them in the Airs.

    If you read through Googles report on the failure rates of all their conventional drives the failures are spread, time wise, across the board. Subsequent discussion referred to the drives becoming cheaper and cheaper. Ergo the failures come earlier and more frequent.

  • by Csound1,

    Csound1 Csound1 Apr 9, 2015 5:31 AM in response to spudnuty
    Level 9 (50,412 points)
    Desktops
    Apr 9, 2015 5:31 AM in response to spudnuty

    spudnuty wrote:

     

    OK I'm answering myself because I want to address everyone.

    So addressing failure rates:

    There is a sudden failure due to power interruption for SSDs. Apparently only the Intel doesn't have this problem.

    Link to this data please.

  • by Csound1,

    Csound1 Csound1 Apr 9, 2015 5:35 AM in response to ThomasB2010
    Level 9 (50,412 points)
    Desktops
    Apr 9, 2015 5:35 AM in response to ThomasB2010

    ThomasB2010 wrote:

     

    Maybe so, but here's an example of catastrophic SSD failure:

     

    http://www.extremetech.com/computing/169124-the-mysteriously-disappearing-drive- are-power-outages-killing-your-ssds

     

    Most HD's fail slowly giving the user time to pull their stuff off.

    So what, I can cherry pick some HDD failures if you want. And one article with one question is hardly evidence, stop spreading FUD

     

    SSD's are more (not less) reliable than HDD's, and anyone who loses data because they don't bother to backup their drive is a fool.

  • by Csound1,

    Csound1 Csound1 Apr 9, 2015 8:58 AM in response to spudnuty
    Level 9 (50,412 points)
    Desktops
    Apr 9, 2015 8:58 AM in response to spudnuty

    spudnuty wrote:

     

    http://www.extremetech.com/computing/173887-ssd-stress-testing-finds-intel-might -be-the-only-reliable-drive-manufacturer

    base paper:

    http://www.hpl.hp.com/people/mark_lillibridge/Power_Faults/final.pdf

    I will try to avoid interrupting the power 1500 times during read/write synchronization. Especially on a Windows machine.

  • by MrJavaDeveloper,

    MrJavaDeveloper MrJavaDeveloper Apr 9, 2015 1:48 PM in response to Csound1
    Level 1 (64 points)
    Apr 9, 2015 1:48 PM in response to Csound1

    Actually, I have to take my own words back about "thousands of power faults" for this to happen. From the link with the actual test report:

     

    https://www.usenix.org/system/files/conference/fast13/fast13-final80.pdf

     

    if you go to section 5.5, we get:

     

    "SSD#3 exhibited an interesting behavior after a small number (8) of tests. SSD#3 has 256 GB of flash mem- ory visible to users, which can store 62,514,774 records in our experimental setting. However, after 8 injected power faults, only 69.5% of all the records can be re- trieved from the device. In other words, 30.5% of the data (72.6 GB) was suddenly lost. When we try to access the device beyond the records we are able to retrieve, the process hangs, and the I/O request never returns until we turn off the power to the device. "

     

    That's only 8 faults before a problem occurred. The report itself is a little confusing because it doesn't really clarify when faults specifically occurred. i.e. it doesn't say SSD 1 ran into it's first problem on test 22 and SSD 9 ran into its first problem on test 1238 (numbers made up as an example).

     

    Additionally, what proof is there that the problems aren't a side effect of the base system they're using for test? For example, if I put an oscilloscope on the power lines going to the SSD, if power is cut what type of transient signal is the SSD getting? Is that unique to the test system hosting the drives?

     

    There are a lot of unanswered questions in the report.

     

  • by spudnuty,

    spudnuty spudnuty Apr 9, 2015 3:22 PM in response to MrJavaDeveloper
    Level 5 (7,097 points)
    Mac OS X
    Apr 9, 2015 3:22 PM in response to MrJavaDeveloper

    "There are a lot of unanswered questions in the report."

    Right, the incredible number of power faults the SSDs were subject to is actually misleading. When one SSD partially fails after 8 cycles what does that say about the probability of possible failure. As more and more people are using SSDs the in use failure rate will become more apparent.

    We've actually seen some failures around here. An SSD failure would catastrophic for me so it just points to the importance of a regular backup policy. 

    I've got a licensed Super Duper so I'll start doing updated backups.

  • by MrJavaDeveloper,

    MrJavaDeveloper MrJavaDeveloper Apr 10, 2015 11:23 AM in response to spudnuty
    Level 1 (64 points)
    Apr 10, 2015 11:23 AM in response to spudnuty

    I'm not trying to say that SSDs are without fault, but I see the following faults with the report as-is:

     

    1. The test uses one and only one test bed. How do they know the problems themselves may or may not be induced by the hosting system itself.
    2. It's not clear specifically what the SSD was doing the very instant the faults were "injected." Was it reading, was it writing, was it idle, was it doing housekeeping? With a multi-tasking operating system at any given instant, even if a device is in a constant READ state (or whatever) the host system may be causing the system and CPU activity to be directed elsewhere in the system, and it's a safe bet that none of the differing SSDs act exactly alike if and when this occurred.
    3. As previously stated, it's not specifically clear exactly where in the tests that most of them had problems, or if they were even evaluated intermittently.

     

    The tests should have been done using a set of different types of host systems. A set of reference systems should have been used that were completely idle as it would have allowed us to know whether or not the problem is truly related to activity or not. The implication is that this only occurs during actual use, but there's really no proof of that. Finally, there's no clear indication of the probability that a failure will occur during the typical lifetime of an event.

     

    I'm not saying these types of problems can't happen, in fact I've read about them before, but in the end the report offers us no correlation to determine the likelihood of such problems occurring, whether it's specific to certain types of systems, or whether it can cause problems for a system that's turned on but not in use.

     

    I will, however, stick with hard drives for backup.

  • by spudnuty,

    spudnuty spudnuty Apr 10, 2015 4:50 PM in response to MrJavaDeveloper
    Level 5 (7,097 points)
    Mac OS X
    Apr 10, 2015 4:50 PM in response to MrJavaDeveloper

    Yes their methodology was somewhat questionable. What we really need to know is when we start up an SSD what the probability of a failure is.

     

    I did find that Google survey on all their conventional drives a while back to be very enlightening. Basically the failure rates are increasing and the expected lifetime is becoming shorter, mostly due to cheaper manufacturing practices.

    "I will, however, stick with hard drives for backup."

     

    I totally agree with that. The "Old Hoss" that you already know most of his quirks especially when he's just about to buck you off!

  • by MrWilliams201,

    MrWilliams201 MrWilliams201 Apr 11, 2015 11:59 AM in response to ThomasB2010
    Level 1 (14 points)
    Apr 11, 2015 11:59 AM in response to ThomasB2010

    I remember seeing a fair number of reports on those types of problems a year or two ago, but I always thought it was limited to OCZ brand SSDs.

  • by ThomasB2010,

    ThomasB2010 ThomasB2010 Apr 12, 2015 11:50 AM in response to MrWilliams201
    Level 1 (13 points)
    Apr 12, 2015 11:50 AM in response to MrWilliams201

    This problem isn't that uncommon. Maybe it's due to power but I know of about 3 people that have had their SSDs essentially reset on them. One minute it's working fine, the next minute all the data is gone.

     

    You might find this hard to believe but some end users have SSDs in their systems and they don't even know it. Some people just take the system as-is from the manufacturer and never understand or find out whats inside.

     

    The normal HDD failure starts out slowly most of the time and people have enough time to get their data off. With SSDs, that possibility may not exist.

  • by HuntsMan75,

    HuntsMan75 HuntsMan75 Apr 13, 2015 11:36 AM in response to ThomasB2010
    Level 1 (14 points)
    Apr 13, 2015 11:36 AM in response to ThomasB2010

    Once again thank you to everyone. I started backing my units up about 3 years ago. Like many, I learned the hard way.

  • by MrWilliams201,

    MrWilliams201 MrWilliams201 Apr 15, 2015 11:12 AM in response to HuntsMan75
    Level 1 (14 points)
    Apr 15, 2015 11:12 AM in response to HuntsMan75

    If you decide to add an HDD to that system along with the SSD, here's what I did:

     

    I needed a partition for Snow Leopard because I have some fairly expensive utilities on it that I still need once in a while, but not ofter enough to spend about $750 on a new set. Originally I set aside a Snow Leopard partition on the HDD and then made the remaining part of the HDD and SSD a Fusion Drive. I didn't like the Fusion because you don't know what file is on what drive at a given instance, plus the Fusion overhead seemed to be eating hundreds of megabytes more. I converted the setup back to an SSD and HDD standalone. The SSD (256G) holds Mavericks, the HDD holds the Snow Leopard partition, a Data partition, and small clone of the active bootable partition on Mavericks. The Data partition holds music, photos, videos, and some shareable applications that can run on both Snow Leopard and Mavericks. Right now the Mavericks SSD partition is using less than 100G of the 256GB on the SSD and it's backup partition on the HDD is 128GB. If my SSD would ever wipe out, I can boot from the Mavericks version on the HDD and be OK. If my HDD ever goes out, I assume I can recover using the SSD. On top of this, I have TIme Machine backups of it all. I can be out in the field and if the system would drop, I likely wouldn't be hurt.

     

    How's that for a system!

  • by ZV137,

    ZV137 ZV137 Apr 16, 2015 10:38 AM in response to MrWilliams201
    Level 1 (54 points)
    Apr 16, 2015 10:38 AM in response to MrWilliams201

    A self redundant system - I love it!!

  • by spudnuty,

    spudnuty spudnuty Apr 16, 2015 10:50 AM in response to MrJavaDeveloper
    Level 5 (7,097 points)
    Mac OS X
    Apr 16, 2015 10:50 AM in response to MrJavaDeveloper

    I always thought this was an interesting alternative to a fusion drive. From OWC blogs:

    Data Splitting

    http://blog.macsales.com/24915-splitting-your-data-an-alternative-to-fusion

Previous Page 2 of 15 last Next