[svlug] More links about ssds

[Rick Moen]

Quoting Greg Lindahl (lindahl at pbm.com):

> On Thu, Jun 02, 2016 at 07:58:23AM -0700, Rick Moen wrote:
> 
> > It's very damn fast for modest-sided writes, but is going to slow down a
> > _lot_ for large ones.  300MB/s is still good, of course -- but it's also
> > going to have a low MTBF because of all the TLC cells.
> 
> It's not useful to talk about MTBF for these devices, because it's
> highly dependent on write activity.

That conclusion doesn't quite follow from its premise.

Yes, the specific figure of MTBF depends on level of write activity, and
more-intense activity means quicker failure and less-intense activity
means more distant failure.  No, that doesn't mean the time before
failure cannot be meaningfully averaged across a set of drives and
compared against that of a different set having had similar write
histories.

Anyway, I believe my point was clear, regardless of what is the most
appropriate 'how long this thing is likely to last' metric.