Spinning rust still typically holds advantages for archival storage, as well.
There was literally a headline on the front page here a few days ago re: data degradation of SSDs during cold storage, as one example.
There was literally a headline on the front page here a few days ago re: data degradation of SSDs during cold storage, as one example.
Or does the data decay there as well, just as a function of time since cells were written?
In other words, is this whole focus on "powered off" just a proxy for "written once" versus "live data with presumed turnover"? Or do the cells really age more rapidly without power?
If you are paranoid about your data and not relying on filesystem level checks from ZFS or Btrfs you should ptobably avoid SSDs for long term storage.
There are "hardware level checks", it's just that they might assume regular usage. If your SSD is turned on regularly (eg. a few hours a day at least), your files are probably fine, even if you never read/scrub your rarely read files. If it is infrequently used, you're right that you probably have to do an end-to-end scan to make sure everything gets checked and possibly re-written.
Why is that? I'd have expected solid-state electronics to last longer at low temperatures.
Or is it precisely that, some near/superconductivity effects causing naughty electrons to escape and wander about?
High capacity hard drives nowadays use heat and strong magnetic fields to write patterns into the platter. It's pretty stable just sitting around doing nothing.
High density multi level NAND involves quantum tunneling a few electrons using a strong electric field into an electrically insulated bit of semiconductor. Over some time the electrons tunnel their way out, but usually this only ends up actually happening if too much writing damaged the insulation.