Jesper Kaas said:
A 1 1/2 year old SSD stopped working recently. Would not boot, but
some files could be read off it.
Can the reason for the early retirement be that I put the PC to sleep
many times a day, whenever leaving it for longer than it takes to make
a cup of tea? I had not heard of "Hybrid sleep", and that was actually
turned on. If I understand it correct, a file the size of installed
RAM (4 GB in this case) will then be saved every time the PC is put to
sleep.
Got a new SSD for free from the supplier, and continue to put the
animal to sleep maybe 5-10 times a day, but have disabled hybrid
sleep. It should be safe then, right?
http://www.zdnet.com/blog/storage/ssds-no-more-reliable-than-hard-drives/1483
http://www.tomshardware.com/reviews/ssd-reliability-failure-rate,2923.html
Flash memory incurs oxide stress on writes hence the need for wear
levelling (
http://en.wikipedia.org/wiki/Wear_levelling) and remapping of
failed memory blocks (just like remapping of bad sectors on magnetic
media). From the Tom's Hardware article, an SSD *should* last 18 years
that experiences 10GB/day of writes. So how many writes in GB does your
SSD experience? Maybe you're incurring 100GB of writes per day so the
lifespan is down to just 1.8 years. It's probably not linear but you
get the idea: the more your write, the more reduced is longevity. 4GB
isn't that much unless, of course, you're power cycling the SSD every
minute which thereafter immediately comes back up (i.e., the power mode
bounces) so you'd be writing 5.8TB every day. You never gave a clue as
to the volume of writes to your SSD but then it's likely you didn't
monitor that volume. However, you do know (or can find out) the
interval at which the low-power hybrid sleep will activate. I don't how
you prepare tea. Perhaps you start with cold water and heat over a wood
fire to bring to a boil and then steep the tea for many minutes to make
it strong. It takes me 5 minutes using a microwave and then add a brew
bag and I like mine a bit weak. 5-minute power mode bouncing (since I
don't know how long you leave the computer in sleep mode) means you're
writing 1.1TB/day just for the hibernate file, and then there is how you
USE your computer that would up the write volume to the SSD. Only you
know or can find out how much write volume you're impinging on that SSD.
Note in the articles that failure isn't just due to the flash memory
itself but also in the interface logic which is using the same type of
components as for HDD interfaces. Stats on HDD failure extend over 50
years whereas stats on SSD failure is just a couple years old.
Obviously the marketing hype is trying to sell the more profitable SSDs
so they are bent to lie, misrepresent, or exaggerate. SSDs are much
faster on reads; however, they are slower than HDDs on writes. The
assumption is that your setup performs much more reads than writes to
exhibit an overall speed increase in using SSDs.
For now, just put the failure to infant mortality unless you know that
your write volume has been huge.