The Anti-SATA Window Protest

Agreed

nah man ise slot.

data is temporary, glory is forever

It will die. There is MTBF but the curve is not linear. You get early faults with any vendor. Get yourself raid5 or equivalent at least if you want to keep the data.

1. @

   ▾ @giomasce · 2y 3 replies ▸

   RAID is not for keeping data safe, but for keeping systems spinning. If you want to keep data safe then you need a backup.

   Report

   1. @

      @SamsonovAnton · 2y

      And you need to verify regularly that the backups you make are actually valid. 👌

      Report
   2. @

      ▾ @CcxCZ · 2y 1 reply ▸

      I mean sure. You better not be YOLO overwriting your important data in a way that can corrupt it irrevocably without having safe copies. But that is tangential to the storage device choice. Offsite backups, offline backups, etc. are certainly an option. RAID >1 is a type of erasure coding and that is certainly applicable to distributed backups too.

      Report

      1. @

         @CcxCZ · 2y

         Also important thing: on HDD you really want multiple copies. Because what the linked studies and whatnot don't cover is read errors - particularly due to magnetic record weakening over time and getting mangled by nearby writes. That's why every raid-like system has check&repair function you are supposed to utilize periodically. And sure, you could set up zfs with two copies on one disk. But at that point you will be much better off with more drives to also tolerate full failure of single drive. And at that point it's pretty much down to your storage and $/TB requirements. Which I haven't seen mentioned at all, but I also haven't read the full backlog.

         Report

That's the thing. For drives dying of age after prolonged use there usually are signs that it needs replacing. For it dying early due to manufacturing faults not so much.

1. @

   @SamsonovAnton · 2y

   No, they usually just die suddenly without correlation to any monitored SMART parameter (with the exception of temperature, which is not a characteristic of the drive itself, though). Age itself is also not a factor to worry about, unless you consider eternity; not to be confused with rewrite count of a solid-state drive.

   Report
2. @

   ▾ @SamsonovAnton · 2y 1 reply ▸

   I once again recommend reading the large-scale study reports mentioned in earlier comment (and the preceding one) to get rid of all popular misconceptions regarding storage device reliability. Those were from HDD era, but I'm pretty sure that most (all?) of the conclusions apply to SSD as well, with the only exception of inevitable wearout.

   Report

   1. @

      @CcxCZ · 2y

      Poor choice of words on my side. Better would be: unlike gradual failure that *may* happen with prolonged use (even HDDs have been doing block relocation for a while now and that will certainly indicate drive failure if if keeps rising), in the early use failures there is not even that indication.

      Report

For 100$ you can only count on hhd. Don't buy cheep chinese trash. Seagate or Toshiba. But remember, it will serve 10-20 years (depending how often use) and one day just crush. So if it important data, just buy 2 data storages, once 1 die, you have a backup already. I payed 200$ for data recovery from hhd, and it is very cheap, price for such procedure can be up to 500$

1. @

   ▾ @CcxCZ · 2y 1 reply ▸

   And the results are very uncertain.

   Report

   1. @

      @Exelution · 2y

      thats for sure... last year of data has erased (

      Report
2. D

   dev_meme · 2y

   For 100$ there’s a good chance of failure even within 3 years timeframe :(

   Report

Fuck SATA All my homie use PCI Express

Stability and overall quality is among first compromised to make it cheaper 🥲