Arch Linux

xy1vro4

Member

Registered: 2025-07-06

Posts: 33

HDD idle or standby (spindown)

For example, if you do a spindown via hdparm
What is better in terms of hard drive wear: stop the disks or let them spin idle? Let's say my disk starts 4-6 times in 24 hours.

the main thing is not to fill up the start cycles, but what about the bearing in the hard drive?

Last edited by xy1vro4 (2025-09-15 00:26:52)

cryptearth

Member

Registered: 2024-02-03

Posts: 2,104

Re: HDD idle or standby (spindown)

this comes close to the one episode of Mythbusters when they tested whether it matters if you keep the lights on or switch it off when you leave the room and switch it back on when you reenter the room

TL;DR: it depends on the drive's specifications if they're designed for 24/7 or not

I have a zfs array and long time used stadndard seagate barracudas - officially seagate markezs them as your every days consumer drive - they're not designed for 24/7 and not for stacking multiple of them in one enclosure - this resulted in a near catastrophic triple failure: one drive failed - while rebuilding a second drive drive failed and after I rebuilt the first drive a third drive failed
luckly I managed to recover thanks to raidZ2 (RAID6 / dual-parity)
the replacement drives are toshiba n300 designed for up to 8 drives in the same rack
and I plan to replace the remaining seagate drives with WD red plus - which are designed for up to 24 drives in the same stack + for diversity

so - look up your drives model and datasheet for thier designed usecase
if you after power consumption: it's negligble as spinning a sleeping drive back up cost a lot power than just keep it spinning - but you benefit from reduced heat and therefore reduced cooling

mpan

Member

Registered: 2012-08-01

Posts: 1,593

Website

Re: HDD idle or standby (spindown)

I second cryptearth in that it will greatly depend on the specific model. Sometimes even batches may differ. There is too many factors to consider.

Focusing on the two most important in 24/7 operation, bearings wear and electronics randomly dying, I feel like both became practically irrelevant nowadays. At least in mechanically stable and well ventilated environments (read: PCs, servers, but not laptops or attached enclosures).

I normally avoid using singular examples as an argument, but since I expect soon this thread will be filled with anecdotal evidence, I’ll share my own HDD. It’s WDC WD1002FAEX-00Y9A0, which has 112,128 hours of almost 24/7 operation on it, and far from being idle. Its twin is 92 kh only because it was disconnected for some time due to a SATA port physical failure. To put this into perspective: the field recording extended test timestamps is 16-bit and soon will overflow for the second time.

---

Paperclips in avatars? | Sometimes I seem a bit harsh — don’t get offended too easily!