> As for SSD specifics pls read
https://wiki.archlinux.org/index.php/SSD> it's a nice overview. As you can see, for SSD, any disk operation is 'bad'
> so you would try to put as much I/O as possible into RAM, for example avoid
> swapping and following Matus' aboce suggestions.
> ('bad' meaning that once all SSD cells are filled, the drive will become
> much slower. You may have noted the same effect with USB pen drives / memory
> sticks or SD cards.)
It seems you're confusing wear out and the slow down seen when the drive
is full.
The technology used in Flash memory "wears out" after some number of
writes. That's true. But nowadays it's only a problem that SSD drive
manufacturers have to solve. End users aren't affected
nowadays, really.
The technology used by drive manufacturers (called FTL) to solve the
above problem has performance implications which typically result in
slower operations when as the disk becomes "full" (for some definition
of "full" as seen from the disk itself, which fundamentally has no idea
what filesystem is used and whether a block belongs to a file or not).
But here as well, this shouldn't significantly affect modern SSD drives
(typically because the manufacturer doesn't let you use all the memory
on the disk, so it's never really full, and because algorithms have been
improved the avoid pathological cases).
The FTL also incidentally introduces a significant amount of complexity
in the firmware which is a source of bugs. This also affects the
reliability of SSDs. But here, yet again, I haven't heard any horror
stories in quite a while, so it seems that it shouldn't be a real
concern for a modern SSD.
IOW, all in all, a modern SSD can be used "any way you like" without
taking any special measures, just like your old HDD, and you should
expect it to perform well and live a long life. Even if it's "full" and
even if you swap to it like crazy (which won't be the case anyway,
since swap is so rarely used nowadays).
Stefan