My buddy wanted a small, cheap SSD for the boot/system drive in his 24/7 Mythtv box. He uses mechanical drives for storing the actual recordings and videos, so this drive would just hold the boot files, programs etc, and mysql database. Plus daily backups of the mysql database.

So I found him one from the pile here, a demo unit I received from the first generation of Indilinx based drives: OCZ Vertex 60GB, with MLC flash. About four (?) years old, formerly used in my own Mythtv box for the same purpose.

A month later, my buddy rings up in a panic: dead!

Turns out I made a tiny error setting up the crontab entry for the mysql backups.. rather than once a day, it was doing them once a minute.. oops.

That's at least 150MBytes of churn, more likely 500MBytes with mysql inefficiencies, once a minute, continuously for about 35 days. 7.5TB - 25TB. On top of "normal" system activity.

So, if you want to kill a small, used SSD, there's a good recipe!

On the plus side, despite the incredible amount of I/O this generated, neither of us actually noticed any kind of slowdown due to all of that I/O. SSDs rule!

I also have a corporate client who uses lots and lots of mechanical drives in some Big Iron. They once had a similar issue, with an operating system updating the filesystem "superblock" too often. Wore away the oxide coatings on that region of the drives!

Thinking about this a little more, I suppose those numbers get doubled again due to the filesystem journalling.. and that's all before the write-amplification factor of the SSD itself.

Is there hope for inventing solid state storage technology that doesn't have a re-use limit for each bit?

Sure there's hope. But right now it's already a lot better than mechanical media for nearly all purposes.

I suspect the problem was overheating more than wearing out the flash, but dunno.. totally dead now.

Wow, that's pretty impressive!

Yes, ST-MRAM is coming in the next year, parts sampling now. ST-MRAM is basically non-volatile DDR3, the holy grail. Density is small now, but it will increase in time.

http://www.mram-info.com/everspin-offici...-available-2013

Coooooooool.

Holy Grail indeed. Thanks for the link.

Or perhaps not.

I've dug out my old notes from back when this SSD was state of the art. After attaching a jumper to the "factory" pins, it came up in "factory mode". A bit of searching around, and I found the latest v1.7 "destructive firmware update" for the unit, which works in "factory mode".

Update applied, drive is now alive again.
It was running v1.6 before the crash, now has v1.7 on it.
Dunno what the differences are.

I don't expect a huge amount of life is left in the drive, but it's probably still got 20-30% in it yet.

Cheers

Personally, I'd want to see a changelog entry that says "Fixed: drive randomly drops dead" before putting that unit back into use...

Peter

Mmm.. apparently my problem was I didn't get the SSD flash memory hot enough when trying to fry it.