I keep hearing people say that hard drives won’t last long and to always have backups. But if it is like that, that means you would have to be buying drives consistently? Has anyone ever had a hard drive work for them successfully for a decade or even more where they wouldn’t have to be buying more?
8 years mark with my WD Red 3TB drives, still zero reallocated sector count. They were kept always spinning, I don’t know if this influenced things for the better
40mb wd had drive in an amiga 1200, still works fine… must say its not had much use in recent years. Think it was bought in 1993
I bought maybe 100-120 HDDs of all brands and capacities since ~2001 and I believe 3 of them have failed so far:
- One 2 TB WD Black from ~2010 died after 3-4 years (one out of 20)
- One 2 TB WD Red from ~2012 died in 2021 (one out of 10)
- One 8 TB Seagate Exos (NLSAS) from ~2018 died recently (one out of 10)All others are still spinning or have been retired due to their small size. In particular, eight of the 2 TB Reds are still in the server which is sitting in a dusty, wet, and cold garage. As I said, one has died, and one is a cold spare that has 0 hours despite being made more than 10 years ago.
I have a hard drive from 2008 that’s fine.
HDDs usually die with a U pattern - they either die very quickly, or after a very, very long time. There’s plenty of working decades-old HDDs.
so if it doesnt die within the first year it will probably last close to a decade?
does desktop HDD last longer than laptops?
I have a drive (wd green 3tb) that is now in like 12 years, and has a uptime of 7.5 years. It works but for me its in EOL hahah
I have 2x 1TB drives from 2010 that are still functional, though I don’t actually use them for anything important. They keep the latest disk image backup for a workstation, but the images are already backed up to a NAS.
I have a few IDE drives around for some of my experimental/ project pcs. 40gb or so.
I have also had to replace a 500gb or two in the raid5 array I use for main storage on my primary NAS.
I try to avoid buying things like 20tb drives on black Friday because if a drive is going to fail, it probably won’t be while there is a sale going on and I do like have a spare drive on hand.
I did pick up two 2tb drives yesterday, but that is because I am finally at a point where I could replace a 2tb drive in an emergency. I still plan to run them mirrored.
My oldest HDD in my main tower now has 40k Power on Hours and around 5k power on cycles, i think it’s around 15 years old
Yes, the longest serving ones I’ve ever owned were from Hitachi. I’ve owned the Deskstar models from 1, 2, and 3TB capacities, which most were purchased as “gen 1” releases. All disks were in operation 24x7. Of them, from memory, looking at my purchase history and notes, are as follows:
- 2x 1TB - one purchased early 2007 on launch, and the other in 2008 - both were replaced with larger capacity disks before they died, probably between 2016-2017.
- 2x 2TB - one purchased in 2011 and the other in 2012 - one died on July 15, 2021, and the other was retired from active service in February 2023 for a Seagate Exos X18 18TB. The retired disk was used from May 2012 to February 2023 for mapped Windows Libraries and secondary program installations to offload the storage from the SSD boot drive.
- 2x 3TB - both purchased in April 2011 for use in a JBOD configuration with a QNAP TS-212 initially. In 2015, they were migrated to a TS-453A. In February 2018, one of them indicated abnormal sectors and were replaced with newer Seagate 10TB NAS disks in a proper RAID configuration. The non-errored disk was donated in late 2021 to my nephew in-law to use as a storage disk for games where it’s still in use today.
Now for some more gory details that might make some people here uncomfortable.
-
!All the disks were purchased in the US initially and used within the US. !<
-
!The first 1TB disk purchased in 2007 was used in an external powered HDD enclosure for transporting FRAPS footage and other media related assets back and forth from uni. It was somewhat protected with a padded cushion lining the bottom of my backpack. After it served it’s purpose, it was pulled from the enclosure and added into my tower with the 2nd newly purchased 1TB disk.!<
-
!Both the 1TB and 2TB disks were shipped via UPS in the Spring of 2013 from the US to Germany along with my Desktop PC and accessories. The HDD cages with the disks inside were pulled from the NZXT Switch 810 chassis, wrapped in layers of bubble wrap, and packed into the same box as the tower, surrounded by tons of packing peanuts. The chassis got damaged by UPS during the shipment with the left door being too bent to be put back on, however the HDDs were fully functional. !<
-
!Both the 3TB HDDs within the TS-212 were wrapped in a Corsair PSU black drawstring bag and placed, feet down, into my Samsonite soft shell carryon bag for my flight to Germany. !<
-
!All the disks did a pretty good job under the circumstances including multiple moves within Germany over the next couple years, and while being in use nearly 24x7.!<
I have a 200MB Seagate coming from the 90s that still works fine and it was untouched from 2001 to 2019. Yes, I had to buy MANY, MANY, MANY drives in the meantime, even if that drive didn’t die.
When I started my first serious networking job, there was a syslog server in our datacentre that had been running nonstop for almost a decade. It was an ancient radiator-white supemicro 3u server with 6 SCSI disks. I decommissioned that server 7 years later. Those SCSI disks had been running nonstop for 16+ years without a single problem. The inside of the server was covered in black plastic dust from the slowly disintegrating case fans. Other than half the case fans not working, there was nothing wrong with that server.
I’ve more than once seen a scenario of a 386 or 486 box somewhere in the corner of a server closet that has been running untouched and uninterrupted since the mid-90s, performing some absolutely critical process, with no one in the company knowing exactly what it is. Everyone who could’ve possibly had a clue has retired decades ago.
The only consensus is to never touch it.
This is more common than many people imagine. And it’s a ticking timebomb.
However, it also speaks volumes of the sheer quality of old-school hardware (and software). Most modern stuff has to be replaced (/rewritten) every few years. But there is more COBOL code running untouched from 3 human generations ago that our entire societies depend on than most people would be comfortable with.
https://www.theregister.com/2001/04/12/missing_novell_server_discovered_after
This machine kept working but was missing for 4 years. They traced the network cable and found it got buried behind a wall but it was still working.