2

When a hard disk is aged and near death, I've experienced unusual noises or intermittent operating system booting or utter failure to boot. In my experience when a hard disk is having problems death is imminent so I think it's prudent to replace a hard disk after about 3 or 4 years. I would backup a hard disk a few times a week to a Flash drive selected from a set that is cycled because I like having multiple backups because if there's a problem with the one that's 12 hours old on one physical device, I can use the one that's 36 hours old on another physical device. Since I believe Flash drives have firmware, it feels risky, so I have multiple Flash drives that I cycle through. I don't backup programs so my backups always fit on a Flash drive.

I haven't tried continuous backups but I would probably still replace a hard disk after 3 years even if I had a continuous backup. I kind of dislike continuous backups because I would probably have only one external hard drive to do it so that seems risky. Furthermore backing up a rotating disk with another rotating disk also seems risky.

What is it like when an SSD is old or near death and if that SSD is in my main system what is the prudent plan to deal with it?

  • With HDDs or SSDs, you can get errors seen in chkdsk. Both should have SMART data too which can indicate issues. SSDs are considered less reliable. And they won't make sounds. – barlop Oct 13 '15 at 4:22
  • If you're doing extensive backups, it seems like a waste to toss perfectly good drives after 3 or 4 years. They can last way longer than that. If you are tossing them while they're still in good working condition, save them as emergency drives and use the active ones longer. A HDD may give you more warning than an SSD; you may get early symptoms (not guaranteed). SSDs tend to handle problems until they can't. SMART data is important to track for both. – fixer1234 Oct 13 '15 at 4:24
2

Tech Report tried to kill a few drives, and the results were interesting. I'm combining data from there and ssdendurancetest.com in my answer.

There's no 'expected' standard failure mode for a old SSD.

Best case scenario would be that the drive goes into read only mode. Intel drives do this, and chances are you can get your data out should that happen. Apparently one of the drives that ssdendurancetest tested throttles back on errors. It was a 'premature' failure, but its indicitive of what may happen with the drive.

Most drives tend to keep on running until they die and it tends to be sudden. What should you be looking out for?

Reallocated sector counts seem to be a good sign that a drive is starting to fail. That said, TR's limited sample size only really started to fail at 100TB at worst (and that drive ran for another 800TB) and their best case drive ran fine until 1100TB, and failed 100tb later.

However if you suffer from something other than age related flash failure, there may be no warning.

Continuous backups make sense since unless you're seriously unlucky and everything fries at once. Backups mitigate risk, not eliminate it. It takes pretty bad luck to lose your main and backup drives at once, and well, its better to have a backup, than to not have one, lose your drive, and then realise there's nowhere to get the data from.

  • If there is a physical disaster and the cable between the main drive and backup drive is short, the physical disaster can affect both. – H2ONaCl Oct 13 '15 at 5:31
  • My answer covers that. You still have fewer failure modes. I have 3 distributed backups around my apartment, and I'm still hosed should my house catch fire. I've survived the loss of up to 2 backups at once tho. – Journeyman Geek Oct 13 '15 at 5:31
2

With SSD's there are not usually any warning sounds (ie slow reads, noises) which alert you to the fact that it is dying. Unfortunately - and contrary to what one would expect - when SSD's fail they typically do so suddenly and catastrophically. (The good news is they are an order of magnitude more reliable then HDD's - but this does not help you if your data dies, so back up often - and if you are worried about this specific mode of failure use RAID.

I do note that SSD's often have a WEAR level indicator as a SMART attribute which gives you an indication of the expected life of the drive, so you may be able to monitor this to give you some indication of gradual failure - that said, this is not always implemented and is implemented differently on different drives, so you need to research before you purchase the drive.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.