SSD Problems

[BradHatter]

I thought SSDs were supposed to be problem free!!!

I've been battling periodic lock ups using my system. I finally pulled out a copy of Scannerz and did a test on it and it found a bad block. The only reason I ever got Scannerz was about a year ago I was having problems with my then 5 year old drive. It was toast. It had almost 10GB of bad data on it probably from a head crash. I pulled it and put in an OWC SSD because I thought these things weren't supposed to have problems until they're like 5 years old. Seemed reasonable.

I called the Scannerz people and they told me I could either use Phoenix to copy the data off the drive onto an HD, wipe the SSD and then clone the data back onto the SSD. They said Phoenix would flag any files that failed to be copied in the log files and that way I would know what to replace from back up. The other option they gave me was to completely re-initialize the SSD and restore from a back up. They said the SSD should correct itself using either of these techniques and if it doesn't I should contact OWC for a replacement.

The thing that's ticking me off is that it seems I'm back to being reliant on HDs YET AGAIN!!! I'm using HDs to clone to and restore from and it's time consuming. I used the old OEM drive in the system for 5 years without a problem. Here I am, just about a year later, and an SSD is basically giving me the same types of problems.

Why did the SSD leave a bad block on it and why didn't it correct it. I thought these things could automatically fix these things. If given an option to get a replacement SSD or an HD instead, should I opt for an HD instead? Am I just unlucky?

 

[yjchua95]

I thought SSDs were supposed to be problem free!!!

I've been battling periodic lock ups using my system. I finally pulled out a copy of Scannerz and did a test on it and it found a bad block. The only reason I ever got Scannerz was about a year ago I was having problems with my then 5 year old drive. It was toast. It had almost 10GB of bad data on it probably from a head crash. I pulled it and put in an OWC SSD because I thought these things weren't supposed to have problems until they're like 5 years old. Seemed reasonable.

I called the Scannerz people and they told me I could either use Phoenix to copy the data off the drive onto an HD, wipe the SSD and then clone the data back onto the SSD. They said Phoenix would flag any files that failed to be copied in the log files and that way I would know what to replace from back up. The other option they gave me was to completely re-initialize the SSD and restore from a back up. They said the SSD should correct itself using either of these techniques and if it doesn't I should contact OWC for a replacement.

The thing that's ticking me off is that it seems I'm back to being reliant on HDs YET AGAIN!!! I'm using HDs to clone to and restore from and it's time consuming. I used the old OEM drive in the system for 5 years without a problem. Here I am, just about a year later, and an SSD is basically giving me the same types of problems.

Why did the SSD leave a bad block on it and why didn't it correct it. I thought these things could automatically fix these things. If given an option to get a replacement SSD or an HD instead, should I opt for an HD instead? Am I just unlucky?

You just got galactically unlucky. I've no problems with any of my SSDs.

 

[Samuelsan2001]

This

You just got galactically unlucky. I've no problems with any of my SSDs.

I agree with this, as with any electronic equipment SSD's can fail, they are very reliable and long lasting in general but they aren't perfect.

 

[Gav Mack]

OS X needs trim enabling on non factory drives using Trim enabler, has to be reactivated upon every OS X point and OS upgrade as Apple disable it by default for third party storage devices. Once that is active then OS X can manage the drive correctly.

You've been unlucky, last crucial or Samsung SSD I had go down on me was a Crucial M4 over 18 months ago and there's been an awful lot since.

 

[BradHatter]

Thanks guys! Greatly appreciated.

So, here are my new questions:

If I do as the Scannerz guys say, will this problem disappear? They seem to think that once the controlling card on the SSD actually detects the problem it will correct it. Is this true?

Why didn't the controlling card pick it up? They told me that some of the clean up routines aren't instant. It was just waiting for the right time to clean it up and then it would fix it.

What about this Apple-trim issue? Do I need to do something special to make sure this things correcting itself or do I need to do something else.

You guys probably think I'm a dummy. I like to think of myself as semi-technical. Maybe that means I know just about enough to get myself in trouble!

 

[2984839]

SSDs absolutely can fail. My Kingston is on its way out as we speak.

If you still want an SSD, any of the big name brands should be reliable. The new Samsung 850 Pro has a 10 year warranty, so that's what I'm going with to replace my Kingston.

 

[Samuelsan2001]

Some general answers

Thanks guys! Greatly appreciated.

So, here are my new questions:

If I do as the Scannerz guys say, will this problem disappear? They seem to think that once the controlling card on the SSD actually detects the problem it will correct it. Is this true?

Why didn't the controlling card pick it up? They told me that some of the clean up routines aren't instant. It was just waiting for the right time to clean it up and then it would fix it.

What about this Apple-trim issue? Do I need to do something special to make sure this things correcting itself or do I need to do something else.

You guys probably think I'm a dummy. I like to think of myself as semi-technical. Maybe that means I know just about enough to get myself in trouble!

Yes the garbage collection and organisation firmware could well fix it.

As they said the times when the SSD can self clean are governed by when it is in little or no use.

You can download Trim Enabler to allow your OSX to perform TRIM on third party SSD's go here.

http://www.cindori.org/software/trimenabler/

Nope you are the same as us all and just have to accept that you learn these things over time, usually in response to a problem.....

 

[FrtzPeter]

Well, in the event that you decide to go for more storage space and a proven track record of reliability with a hard drive, I'd be more than happy to recommend this:

HGST Travelstar 500GB 7,200 RPM SATA III 6Gb/s 2.5" Notebook Hard Drive PN 0S03788

The new Hitachi's (it has to be the new line, though) have something like 3-4 times the data density on platters, lower power, and a rotational speed increase of 7.2/5.4 ratio. Put this all together and you end up with a hard drive that's able to read and write data at almost 3-4 times the speed of your original HD, maybe even higher.

There not as fast as SSDs but they're a lot faster than almost all old HDs, they're cheap costing $50-$60 at most places, and you get 500GB of space.

I've heard of lots of people having problems with SSDs, especially bad blocks. It's actually normal for a bad block to develop in an SSD. What isn't normal is when they can't be cleaned up. When that occurs, it's probably time to contact the vendor.

 

[FrtzPeter]

I was just perusing this site and noticed there seem to be a fair number of SSD problems cropping up.

Could someone please tell me what the signs of an SSD going bad are? For some reason this seems to be topic that's ill defined.

I understand the bad block that the OP had, but what happens to them when they start becoming write depleted? As more and more blocks get lost, does the size of the drive go down? Do they turn into read only devices like USB flash drives? What happens to them?

 

[MacRobert10]

For bad blocks they get to the bad block and it acts just like a bad block on an HD, except no noise. You get I/O errors.

I would think the size of the SSD decreases as the blocks get marked bad, but I can't confirm that.

One big problem used to be the SSD just wiping itself clean, like it reset all bytes to 0, but I haven't seen that in some time and it wasn't age related.

Other than that, I don't know what they really act like when they become depleted. All I'd say is make sure your data is backed up to an HD whether you like it or not.

 

[TheBSDGuy]

Most SSDs that fail, fail catastrophically, meaning the drive suddenly no longer is seen by the system. Whether it's holding data and the data can be recovered, I don't know. I don't think they've been around long enough or in use by enough people for anyone to really say for certain what will happen when the blocks start getting depleted. I'd also think it will vary considerably from manufacturer to manufacturer because of the firmware differences.

 

[Weaselboy]

I was just perusing this site and noticed there seem to be a fair number of SSD problems cropping up.

Could someone please tell me what the signs of an SSD going bad are? For some reason this seems to be topic that's ill defined.

I understand the bad block that the OP had, but what happens to them when they start becoming write depleted? As more and more blocks get lost, does the size of the drive go down? Do they turn into read only devices like USB flash drives? What happens to them?

There is a good article here where they run a SSD right into the ground and document what happens as it started "wearing out". Interesting read.

Most SSDs that fail, fail catastrophically, meaning the drive suddenly no longer is seen by the system.

I agree... that is what I have seen also. You don't seem to get errors or slowdowns like when a HDD is dying. With an SSD it seems like things are going great, then all of a sudden BAM... dead.

 

[Fishrrman]

Frtz wrote above:
[[ Could someone please tell me what the signs of an SSD going bad are? For some reason this seems to be topic that's ill defined. ]]

From what I've read, there probably won't be any overt "signs" of an "SSD going bad".

It will just fail, period.
Probably no warnings, of any kind.
One moment it will be there, and the next.... just gone.

This is why you MUST keep a BOOTABLE BACKUP if you use an SSD as your boot device.

It will literally "save your behind" in a moment of extreme need.

Either CarbonCopyCloner or SuperDuper! can do this for you.

 

[TheBSDGuy]

There is a good article here where they run a SSD right into the ground and document what happens as it started "wearing out". Interesting read.



I agree... that is what I have seen also. You don't seem to get errors or slowdowns like when a HDD is dying. With an SSD it seems like things are going great, then all of a sudden BAM... dead.

That's a good article. Too bad it wasn't done using a number of different manufacturers and technologies.

One point I might make an exception to in that article is that they don't really seem to clarify the difference between a regular bad block (that can be relocated) and uncorrectable blocks (those that can no longer be relocated because the spares are depleted.) Maybe it's just me reading too much into the article, and maybe there's more info on some of their tests.

It's not abnormal for a bad blocks to occur on an SSD over time. I've seen it twice, one on an SSD I own and one on an iPhone. With the SSD, when the system got around to "cleaning up" after itself it went from existing to non-existing. I checked it throughout a day using cursory mode in Scannerz and it stayed in place....then voila, like magic, it just disappeared.

The iPhone was a little more interesting. Although not formally an SSD it is flash memory. With mine it was in the middle of a song. I'd be playing iTunes and at the exact same location every time while playing this song, iTunes would stop, freeze momentarily and just quit. The system would go back to it's home screen like nothing ever happened.

To get rid of this, I re-initialized the iPhone, like it was a "never been used" type of device and then did a restore. To the best of my knowledge the iPhone's system management routines fixed the problem...but who's to be certain? To the best of my knowledge no one makes any tools to test these things. I guess it's because so many people see them as throw-away devices.

 

[Weaselboy]

One point I might make an exception to in that article is that they don't really seem to clarify the difference between a regular bad block (that can be relocated) and uncorrectable blocks (those that can no longer be relocated because the spares are depleted.) Maybe it's just me reading too much into the article, and maybe there's more info on some of their tests.

I think they do kind of allude to it in this section. As I understand what they are saying up until this point any blocks that were sketchy were just marked as do not use and replaced with a "spare" block from reserve. So on May 26 the drive ran out of "reserve blocks" and then began a slow descent to its death.

The first uncorrectable error came four days later on May 26, after 3,187 cycles and 764 TiB of written data.

 

[simonsi]

I think they do kind of allude to it in this section. As I understand what they are saying up until this point any blocks that were sketchy were just marked as do not use and replaced with a "spare" block from reserve. So on May 26 the drive ran out of "reserve blocks" and then began a slow descent to its death.

Any drive (HDD or SSD), will have a finite number of spare blocks, is this behaviour any different to a regular HDD I wonder?? I expect no-one has run a regular HDD to death in testing (outside the manufacturers), in the same way as due to the much lower data transfer speeds meaning such a test may not be feasible....or the drive physically dies first?

 

[TheBSDGuy]

Both HDs and SSDs have spare blocks, but an SSD has a lot more of them. Any HD can have a bad sector crop up here and there over time, but it's really sort of an oddity for it to happen "naturally." Most bad sectors on a drive are caused by head crashes or the drive wearing out to the point the heads start crashing on the drive frequently. This is when the drive is basically shot.

An SSD has a lot more spares because it's normal for them to go bad over time.

With respect to the OP, I think the way most people deal with SSDs is to use an SSD as an internal drive or part of a Fusion Drive and then use a regular HD for back up. Keep in mind that you can get a good HD with lots of space for tens of dollars, not hundreds of dollars.

 

[FrtzPeter]

When you can get an SSD with 500GB of space for under $70, let me know:

http://www.amazon.com/HGST-Travelstar-2-5-Inch-Internal-0S03788/dp/B00JQYTXHC

I had a smallish SSD in my system and it was just like living a decade or two ago when people were using floppies and optical drives to transfer data. I had to keep moving stuff off and managing everything. I probably wasted more time having to compensate for the lack of drive space on the SSD than it saved me.

If you do a lot of intensive disk work, an SSD is worth having. If you don't, the only thing it gives you is fast boot times - after that you'll rarely notice the difference. OS X caches all it's programs in RAM so when you re-start or re-open them once pre-loaded, it never even touches the drive, which is why it's fast.

And now we seem to be reading more and more problems with SSDs.

They're not ready for prime time and they cost a fortune.

 

[lavrishevo]

I got the Samsung 840 pro. Love it. Supper fast.

 

[MacRobert10]

So if they can now develop a disk with 3X the density for an HD it means 3X the speed, right?

If they can develop that over the last 2 or 3 years it would tell me that possibly HDs themselves might be able to compete with SSDs sometime in the future. Basically it sounds like a new Hitachi would have data rates pushing the limits of SATA I, so if they got more density on it they could approach SATA III.

Maybe I'm just fantasizing about having a big disk that's real fast for just tens of dollars.

 

[simonsi]

So if they can now develop a disk with 3X the density for an HD it means 3X the speed, right?

If they can develop that over the last 2 or 3 years it would tell me that possibly HDs themselves might be able to compete with SSDs sometime in the future. Basically it sounds like a new Hitachi would have data rates pushing the limits of SATA I, so if they got more density on it they could approach SATA III.

Maybe I'm just fantasizing about having a big disk that's real fast for just tens of dollars.

Depends how the different density is achieved. If the capacity of a given drive has doubled then it is possible that has been achieved by having 1.4x the number of tracks and 1.4x the amount of data on each track.

The read/write speed of the disk comes from the rotational speed and the amount of data stored on a track (the circular data string under one head). So in this example a drive of twice the capacity may have 1.4x the raw read/write speed (ignoring cache and some other effects.)

However in real-world user experience it is the seek time (the delay in moving the heads to the new track where the data is), that is most of the time to get the data into RAM, in this the SSD will always beat the HDD hands down as there is no physical movement involved.

 

[ZVH]

Seek times, in milliseconds, will typically only occur a few times during a read or write because that's what it takes for the drive heads to position themselves. The OS is supposed to use the file system to optimize sequential reads and writes and minimize the movement of heads from track to track. The primary bottleneck with HDs compared to SSD's is thus the read and write transfer rates from and to the surfaces of the drive. Because of seek times, an HD would never achieve the performance of an SSD, but it could theoretically approach it.

If someone could build an HD that could actually transfer data to and from the platters at SATA rates, yielding transfer rates not as fast as SSDs but possibly approaching low-end SSDs, considering costs it would probably threaten SSD sales somewhat. Few people really need the speed of SSDs, IMHO, but in todays age of mega-sized documents, people do need storage space.

 

[simonsi]

The primary bottleneck with HDs compared to SSD's is thus the read and write transfer rates from and to the surfaces of the drive.

Really? Have a read here, Random Read Latency is the key driver in real-world OS usage as that constitutes most disk access by both OS and users. In that regard SSDs are some 20-30x faster than HDDs, any HDD will never approach an SSD for that performance given the laws of physics.

http://ocz.com/consumer/ssd-guide/ssd-vs-hdd

Note the "Faster" section on both RRW and some App launch figures. In most random small file operations the SSD will have completed the task before the HDD has completed its seek to the correct track...

 

[ZVH]

Really? Have a read here, Random Read Latency is the key driver in real-world OS usage as that constitutes most disk access by both OS and users. In that regard SSDs are some 20-30x faster than HDDs, any HDD will never approach an SSD for that performance given the laws of physics.

http://ocz.com/consumer/ssd-guide/ssd-vs-hdd

Note the "Faster" section on both RRW and some App launch figures. In most random small file operations the SSD will have completed the task before the HDD has completed its seek to the correct track...

Believe it or not, I don't disagree with that, not at all. There's no doubt in my mind that an SSD will be faster than an HD. However, unless someone is using an application that requires continuous access to data over fairly long periods of time, the difference may not be noticeable to end users.

Open up a copy of Activity Monitor, Performance Probe, or whatever it is that you want to monitor your system with, and look at the disk activity that's going on. Basically, once the system is booted and applications are loaded, the drive really doesn't get hit all that much any more. I can't speak for Windows in this issue, because I don't use Windows. Most people don't run applications like databases that are continuously being hit by tons of users on their systems.

Considering the way HFS works, which is by using in-line optimization of discarded blocks, it will cut down on seeks, making the surface to controller data rates that much more important. Fragmentation, IMHO, really doesn't exist with OS X, or at least not much, and of course applications that tell you that you need to use defragmentation to optimize and SSD are just flat out lying to consumers.

An HD will never equal the performance levels of an SSD, but in normal user circumstances, I think it's quite possible that, given possible advances in HD surface technology, they may actually be able to still give SSD's a run for their money, at least for the typical end user anyway.

 

[FrtzPeter]

If anyone is interested, here are the boot speeds I got for the old Fujitsu OEM drive, the new Hitachi that I linked previously, and the SSD I had:


Fujitsu: 1 min, 15 sec
Hitachi: 47 seconds
Sandisk SSD: 30 sec

That's with Mavericks.

And now for some irony:

I have an old 2004 Aluminum PowerBook that boots Leopard in about 30 seconds.