Re: SSD secure erase

Fabulous. So you find a 10-year-old SSD in the back of a server room. You don't know what's on it, but it could well be financial details. It doesn't boot up, the drive controller doesn't recognise it at all, and there's no buttons/resets on the drive whatsoever. A data recovery firm will charge you several thousand pounds just to get the drive back to operational to find out what's on it or issue a secure erase command (if that's even possible by then).

ATA secure erase commands predate SSD's by about a decade. There's a reason you don't rely on them. When a drive dies, you still need to destroy the data, but you can't make it do it itself. You tend not to replace drives unless they've died or you don't want to waste time cleaning them up for re-use. And waiting hours while each drive of 1000's is powered on, issued an erase command and waiting for its successful completion (which could still take hours) is the least efficient way possible to destroy the data (the most efficient being to throw things in a fire until there's no recognisable components left).

Re: I've gone for large capacity.

It depends.

The number of rewrites is only given "per cell, on average", which is very different to the real-world maximum. You might find that it writes the same cell over and over (even with wear-levelling) and burns through the replacement "spare" cells too, writing that same information, and you end up with a worn "hole" in the disk even though millions of other cells are untouched.

The wear-levelling algorithms are not perfect, and the averages given mean that any write could fail and bring in spare cells to replace a broken one, regular or not. It really depends on how much "spare" capacity you have to be able to bring online should a cell fail more than anything else. With 500Gb, you're likely to have less spare (proportionally) than a 100Gb.

I'd actually care more about just using the drive normally and choosing a manufacturer/model with a good reputation than any advertised statistic, though. Base it on real-life experiences, buy the "older" model with more good reviews and less "it just died on me", which would tend to be the smaller one, and would also tend to cost more than equivalents of its size (even to the point where you could get a 256Gb for the price of a particularly reliable 128Gb).

Or, don't use it for permanent storage and just realise you're going to kill it like any other drive. Have a spinning disk for actual data, and an SSD for "working storage" (e.g. games, Windows, things that can be replaced and don't actually matter).

People keep telling me that spinning disks have a lifetime and die. The only disk I've personally witnessed dying without giving LOTS of warning before hand (e.g. SMART etc.) was actually 20Mb. Ever since then, I've seen disks that report 1-2 bad sectors and you can just rewrite those sectors and get around it (either the disk pulls in spare sectors, or the OS marks them as bad and doesn't try to use them), or they just keep going. Hell, I have a stack of disks at home that go back to 386's.

Buy it, expect it to die, don't rely on maths based on adverts to base your averages on, back up anything critical on a couple of disks of differing technology to cover your ass.

SSD's are really just consumer items now. You're probably NOT going to hit the write limits at all in the lifetime of the drive but, like with any consumer item, it's possible it could blow up on the first day.