WD Red NAS drives and ZFS

[eccerr0r]

Fortunately wear related failure of SSDs are pretty well gauged at this point, at least the ones that keep track of erase cycles for each block which any self respecting company will do. You won't likely find suddenly one day you must replace the SSD due to wear failure.

This is different than other failures like random or infant mortality failures, but HDDs have the same problem anyway.

But agreed the cost per bit is still too high for me, which is the sole reason for not switching to all SSD. Plus scratch disks* were meant to be scratch disks that one keeps writing on...

---

Incidentally speaking about scratch disks, someone handed me a 5¼" floppy disk with a label "Scratch Disk" affixed to it. Noticing it had bad sectors, I spun the mylar around to see if I could see physical damage on the surface.

Turns out, someone actually took a fingernail and deliberately scratched the mylar for the full length of the head access hole for a few times... Wow. *facepalm*.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

[Zucca]

My storage requirements are pretty low. Around 2TB, but with redundancy, so say 4TB (which should leave me some spare storage too) and I could go full SSD on my server, which is the only machine where I still have HDDs.
But I would also need to buy a new server... The current one is quite old, and consumes too much power. I'm thinking of some low power 2U server... Oh well. Too many things going on atm. I need to think that over again maybe next year.
_________________
..: Zucca :..
Gentoo IRC channels reside on Libera.Chat.
--

[pjp]

[eccerr0r]

The manufacturers do stress testing and know that the cells will survive that 500, 1000, 5000, etc. cycles at minimum. You will typically get more if the controller allows it but once you hit that number you've gone out of spec and results aren't guaranteed anymore. As said this wear number is kept track of and you can see it slowly decrementing, so you'll see how many erasures you have left before it's time to replace.

I think if you're planning to use for swap, as long as it's "intelligent" swapping you're fine. Using it as an outright replacement for RAM, i.e., unintelligent swapping, will not work long term. Also note that unintelligent use of swap, it's still not going to suddenly die on you, it's still subject to the same gradual degradation. Also hopefully you're not continually thrashswapping, once you're done with whatever you're building, it'll sit idle and not going to go through erase cycles. I've accidentally thrash swapped ("unintelligent swapping") to my SSDs before and the amount of wear that caused was not a big deal though I do have 2BPC units. One of the machines that I thrash swapped still choked hard, and had to hard reboot it (ran too many threads on the local processor because distcc failed).

Hibernation is absolutely no problem. You're not constantly hibernating, and this is a common use for SSDs.

Using it for PORTAGE_TMPDIR is not that big a deal either. Again you're not constantly merging stuff day in and day out, 365/7...

Again I would think that constantly downloading video torrents and deleting them, or probably continuous full bandwidth surveillance recording would be the harshest workloads, unless you count using it as RAM replacement on a memory bandwidth intensive application.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

[pjp]

You mean manufacturers like WD, Samsung, Crucial?

If speed were more of a priority, I'd have more than just the one in my laptop.

I'm hoping to get back to pruning my redundant copies of data so capacity will at least shift in priority to knowing that I want what I have.
_________________
Quis separabit? Quo animo?