Poll: Have you had a SSD fail? | Hacker News Reader

Poll: Have you had a SSD fail?

23 pointsduck40 comments

Related to Atwood's post about SSD failures (http://news.ycombinator.org/item?id=2505700), thought it would be interesting to gage HN users on their experiences. Please only participate if you actually have/had a SSD for at least 3 months. If 'Yes' leave a comment about drive details and time till failure.

View on HN

I actually had something like 12 SSD failures in the same machine in about 18 months, including 3 of those being an OCZ Vertex 2 and one Intel X25-M. Failures ranged from working fine to completely failing and not even being recognized the next minute (!), to sudden very very slow read/write and drive errors leading to complete failure. Eventually I talked an Apple genius into replacing the logic board in the cursed MBP (second gen unibody, 15", replaced the hard drive instead of opting for the optical bay) and it's been running smooth with a newer Intel X25-M for 4 months now with similar usage patterns.

The first few failures I wanted to chalk up to SSDs being flaky or my being sent refurb drives, but given that I had so many in the same machine and no problems in any other desktops and laptops I tested SSDs with, I just want to blame the machine.

I'm using a RAID0 of two Intel 160 GB multi-level cells. One I bought in the middle of 2009, one at the end of 2009.

Both are under constant use (that raid is the only drive in my development machine) and they run fine so far.

At home I'm running two 256GB multi-level SSDs also from Intel, also in a RAID0. I bought them last December after having one Corsair SSD in a previous RAID0 attempt constantly losing some data (the other one was fine).

3 other people in the office are running SSDs (also 160GB Intel) since the end of 2009. OSes include MacOS X, Ubuntu and Windows XP and now 7.

All work fine so far and I'm not seeing any of the often quoted reliability issues (aside of that one that was defective from day 0 which happens with hard drives too)

And before you ask about the safety of using a RAID0: No I don't have backups but also, no, I don't need backups as all the important files are in git and all of my profile is on Dropbox.

This is off topic, but isn't Dropbox essentially a backup service? Also, having your important files in git isn't really helpful unless you git repo is remotely stored and backup up.

For your sake I hope it is.

If he works with a team, he already has git as a backup. In the event of a drive failure, the others won't have the changes he didn't push but Dropbox probably does.

I had an 80GB X25-MG2 fail within the first week of use (by "fail" I mean a nearly complete irrecoverable loss of data--most probably related to a faulty controller rather than flash chips) . The replacement I got had been chugging along perfectly for a couple of years. I've never had any problems with a Vertex 2 120 drive that I also have.

Both drives had a pretty high read/write usage ratio, though. And most of the rewriting taking place was small in volume. I ordered a Vertex 3 drive last week, and I intend to sync its contents with a magnetic drive in the same system a few times a day, just like I have been doing with the other two drives. So all in all, besides the first Intel drive being a fluke, my experience was quite positive.

I had been under the impression that when SSD drives die, write operations start failing but existing data is still readable. Is this wrong?

Some SSDs react to some failures this way. But not all SSDs, nor all types of failures.

If the failure is not being able to write the block, then all blocks are still readable you just can't write any more. Any other failure not so much. Also note the read may fail at the OS level if it can't modify the last access time.

I had an SSD fail after about a year. It completely ceased being a functioning device. I could not boot my computer at all with the device attached.

But was it mountable as a read-only volume? Keep in mind that most operating systems require a writable volume unless they've been specifically configured otherwise (e.g., Knoppix).

I have never had one fail once it started in production, but have had two X-25M G2 160G drives fail out of the box (one would not respond at all and the other would not recognize more than 8G of available storage.) The higher-than-expected failure rate is annoying, but I would rather have them fail at install/setup time rather than six months later...

In late 2009 I bought 2 Crucial SSDs for my MacBook Pro, replacing a Seagate HDD that had died after just 5 months (the other one went in the optical bay). They're still working fine after 16+ months, and even if they die now they'll have outlived many of my harddrives (75GXP anyone?).

One is my boot drive, and one is my media drive. SMART reports the boot drive has 78% of its erases left, and the media drive 100% (mostly static data). ~20% erases down after 16 months means if the wear leveling works out, my drive should last another 5 years.

My usage is quite normal - coding work doesn't do very many heavy writes. I have 8 GB RAM, so the swap file doesn't see much usage, but instead there's 8 GB of hibernate file writes whenever I put my laptop to sleep (a couple times a day). Also, there's some background process that hits the drive with ~100 KB of writes every couple of seconds.

I posted this on the reddit thread as well, but I've used some SSDs in VM hosts, and seen a pretty damn high failure rate. They're great for VM hosts for obvious reasons, but they thrash pretty hard, and I've taken to triple RAID mirroring them, across different brands.

I've had 2 (n-2 gen) OCZ Vertex SSDs die within days of installing, and 3 Intel X25M-80s that have been rock solid for 6-10 months.

There's nothing like seeing 6 kvm virtual machines boot up in seconds to remind you that they're damn fast on reads.

On my 3rd X25-M. First was a g1 that ceased to be recognized by the OS as a device after 9months, the second was a g2 that had much degraded performance and then started showing many errors, corrupt data and the like so I stopped using it. Just prior it was in a RAID 0 array for a year with another x25m g2 that I pulled apart after degraded perf and had a feeling something like this would happen.I now have a single x25m g2 (the other one from the array) with the OS X trim patch installed.

Waiting for a MBP update to ship with lion and dual sata rev 3 (currently its only one rev 2 and one rev 3) so I can RAID 0 two SF2500 based SSDs like the vertex 3 pro.

Any citation that Lion will rev the latest MBP to SATA 3/3?

It wasn't clear to me that this was a software limitation...

The pull together a slightly larger sample size, consider the reviews of one model of OCZ Vertex 2, widely recognized is a solid SSD, in terms of price, performance, and durability:

http://www.newegg.com/Product/Product.aspx?Item=N82E16820227...

Of the 425 ratings, 17% are 1-star ratings, based almost entirely on a short-term failure measured in < 6 months. That's not to say the Vertex 2 as a whole has a failure rate that high (this is the vocal minority, after all), but it should at least help put some of the nay-sayers in perspective.

I have a Corsair P128 I bought in October of 2009. It cost me $350 but after seeing the performance gains and reading about the almost change in lifestyle you get in using one I thought it would be worth it.

I was right. Everyone was right. It's still running now, without issue, and I enjoy 8 second boot times with Windows 7 and virtually instantaneous-launch applications. Well worth the price. I regularly backup my data but it's nice to know that if it failed today I could buy a replacement from NewEgg for nearly half the price and probably better performance.

I bought a 256GB Crucial RealSSD C300 and put it in my MacBook Pro. It failed catastrophically within about a month. Luckily I do an automated full-disk backup to an external drive every night; between that and CrashPlan, I didn't lose any data, and Amazon gave me a full refund on the drive.

I then bought a 240GB OWC Mercury Extreme Pro and put it in the same MBP. It's been running fine for about eight months now.

One intel X25-M 80GB SSD. According to the SMART information 10% of its write life has been used over 1.5 years. Still going strong in a work machine used day to day.

Two OCZ Vertex 2E's 120GB (separate machines). Had both for 6 months and both are working reliably with no signs of problems or performance degradation.

Not had a single failure so far.

Also had no problems whatsoever with my Intel X-25M, over 1.5 years old now. I've heard a lot of horror stories about other manufacturers, but never met another Intel user who had problems. Supposedly OCZ has gotten much better quality wise, though there are lots of bad reviews for those drives on newegg.

OCZ is mainly using sandforce chipset now... which is is a quantum level leap from the initial jmicron chipset.

A vertex2 or vertexLE is a completely different beast than a vertex1.

I had a Super Talent (UltraDrive GX) fail after 2 days and 2 different OCZ Vertex 2 fail each after about 3 months. All of them failed in my trusty MacBook Pro.

I'm now back to good old hard drives, sure it's WAY slower but the current state of SSDs seemed to be a bit too unreliable.

I suspect that there is something wrong with your "trusty" MacBook Pro.

I have a 3.5" OCZ Vertex2 that's being used on a daily basis (with lots of swapping) since Oktober. No problems as of yet.

I've had another one (2.5") in my Macbook since early january. No problems either.

Performance might have degraded, but not that I've noticed.

I've got a relatively new Vertex 2 which is totally fine, but that's still well within the realm of normalcy. However, I've also got a first gen x25-m, and that is fairly old as far as SSDs go. That one also still works perfectly.

Just had a failure last week on my OWC drive. Replacing it wasn't too bad (had to wait a week) but made me realize that with so much stuff in the cloud, the failure of laptop drives doesn't really bug me or slow me down.

Two X25-Ms over the past two years and neither have failed. But I've been very conservative about which drive I chose due to the possibility. At the moment, Intel makes the only drives with a track record of reliability.

No: I've had my Intel X-25M for over 2 years and its holding up great.

I've had the same drive for almost 2 years, and I have no problems. I have a separate spinning rust drive which I use for my swapfile and for downloads, and of course all my large files are on my fileserver.

I also have had an OCZ Agility 2 in my laptop for about a year now, and that drive hosts a (small) swapfile. No problems.

Both these systems run Linux.

No: I've had my OWC Mercury Extreme Pro for 8 months now and haven't had a single issue.

I've had an Intel X-25M die after about 16 months. BIOS wouldn't recognize it anymore.

Yes, due to firmware bugs.

I've had an OCZ Vertex for a couple of years now (I got it a few months after they came out), and it's been fantastic ever since. I've also got two Intel 60GB SSDs in RAID0 in my laptop, which have worked flawlessly since day 1.

However, I purchased an OCZ Vertex 2 for my parents a few months ago as they were building a new computer. About 2 months went by, then one day -- boom -- computer won't boot anymore. I go over and investigate, and it ultimately turns out there's a bug in the latest (read: shipping) firmware in the Vertex 2 which is tripped by some S3 power state transition (e.g. when the computer goes into hibernate mode) and causes the driver controller itself to stop responding. So, your data is OK, but there's no way to flash the drive or anything because the BIOS won't even recognize the drive anymore.

So, my advice is -- get an SSD (or two, in RAID0, if you can afford it), but make sure to keep any really important files backed up (you should be doing this anyway); also, make sure to update your firmware regularly, since the firmware itself may contain bugs that could wreck your drive.

make sure to update your firmware regularly

Easier said than done with the Vertex 2 if you've got a Mac. The firmware updater is a Windows only program. At least it was when I was installing my Vertex 2.

I have had an OCZ 120GB Vertex 2 for about a month now, and today updated the firmware to 1.33, in the release notes it says reduces the chance of data loss due to power interuption.

I would also advise making regular backups, checking SMART and for firmware updates regularly.

The previous drive I had was a Kingston 64GB and it still works after a year.

I've given up on hibernate for my macbook+vertex2... however, my battery is decent enough to get me through the day and I charge often.

I haven't noticed much of a difference other than the occasional instance of leaving the laptop on and without power (no autosleep for me).

Dropbox+svn/git make sure I don't lose any code.

Have had mine for just over 3 months, no problems whatsoever.

No : Intel X25-M G2 160GB running Windows 7 since 2009-10-28.

Less than a year, no problem.

Reply on news.ycombinator.com