Hey, weird tech #hardware question.

What is the best, most reliable, long term, but still easily accessible, non-cloud storage option for massive gigabytes of data?

Like, I have a xxx terabyte backup drive, but someone recently told me that those aren't reliable long term, unless consistently powered up, as bits can disappear, corrupting files. (It seemed crazy, but other posts seem to confirm it.)

So, if not xternal SSD, are lots and lots of cd/dvds still the most reliable storage media? Surely no?

#datastorage #backups #archives

@MissConstrue

I don't know if optical media is the best choice, either. Bit rot on CD's and DVD's is annoying as sin. Most of the CD's I collected in high school are now fancy looking coasters that I'll probably never be able to bear throwing away.

@helplessduck

Oh no! Really? I have thousands and thousands of CDs I've never burned to another media. OMG, some of them can't be replaced, it's stuff from the soundboards at clubs where I had friends who were sound engineers. Oh man, ok, next week I'll burn those to hard media so at least hopefully there still exists a copy.

@MissConstrue @helplessduck if it’s CD-Rs then most of it is at least damaged, if not lost, by now

@mirabilos @helplessduck

Well, that's a rabbit hole for the list of How To Distract Myself From the Timeline Melting. Hooray? ;)

@MissConstrue @mirabilos

That is one deep warren. I can't find the article, but I recently read about Intel developing a process that uses specialized glass sheets for what they're calling "permanent" data storage. So, if you've got a few million dollars lying around to spend on glass, I'll bet you won't even have to switch to Linux.

@MissConstrue

Oh dear.
Hopefully you'll have very little loss and lots more sense than I did when I chose the Windows Media Audio format for more than 100 CD's. 🫠

@helplessduck @MissConstrue Oh yes. Please back up your unique/rare CDs. Almost all of my CDs from the late 1980's and 1990s have errors, some don't play at all any longer. Of these, some are exceptionally rare small runs, others are one-of-a-kind recordings burned to CDR. I truly wish I had started my archival to hard disk much earlier.

@eric_herman @helplessduck

Yeah, this is stuff like Nirvana when they were still playing to less than 50 people. MotherLoveBone, Jello Biafra, Dead Milkmen, Metallica playing a club in London…just unbelievable things that should be archived for humanity.

That said, I’m sure the engineers have copies too, I doubt I’m the last known copy, but just in case, ya know?

@MissConstrue @helplessduck

I spent several weeks last summer ripping hundreds of CDs. About half were 30-40 year-old commercially produced CDs and I had virtually no issues with them. More problematic were homemade burns, but 90% were just fine. The worst were those magazine CDs from the early 2000s. At least a 1/3 were unreadable.

@MissConstrue HDDs are a good cost‑effective choice for large, accessible long‑term storage. SSDs/NVMe are non‑volatile but NAND cells can lose charge if left unpowered for long periods (worse with high temperature and heavy prior wear); for archival SSD use, power or refresh them periodically (intervals vary commonly months to a year depending on drive and conditions). For multi‑year archives, prefer HDDs or tape/M‑DISC, 1/2
@MissConstrue 2/2 keep multiple copies in different locations, store media cool/dry, and migrate/verify data every few years.
@MissConstrue a live system that is maintained regularly
@jarinks @MissConstrue So if I keep my xxx terabyte hard drive plugged into the laptop, it will be stable for a long time?

@MissConstrue good old magnetic HDDs (not SSDs), and not the SMR ones. The lower the information density, the better, so prefer 3½″ over 2½″ and m.2 is completely out, and prefer 1–2 TB drives over larger ones (just distribute; for long-term storage, speed does not matter much, so don’t RAID 0, at best linear (ā€œRAID -1ā€) but better keep them independent)

CDs/DVDs deteriorate too fast, chemically and also from light. Flash storage… well yeah, no.

I have no opinion either way on magnetic tape, for that you’d need to ask an expert.

For bitrotting (as opposed to sector-damaging) media, might make sense to invest into forward error correction codes (not just checksums).

@MissConstrue I back up to an external 6TB HDD over USB3. It is set to power down when idle and it spins up when asked. I also back up online with BackBlaze, and iCloud backs up my email and photos. So if disaster strikes I’m covered.

@MissConstrue

Agree, as others have already noted, that recordable CD/DVD are probably one of your worst options. (and any sunlight-exposure makes them even worse)

The bit about unpowered backup drives I think is specific to SSDs -- data-center -focused ones supposedly can start to bit-rot in less than a year if left unpowered, others have longer data lifetimes but still measured in single-digit years.

Spinning-rust HDDs are more reliable without power, but still subject to gradual loss (as well as mechanical failure). Best bet with HDDs is some kind of array like ZFS that does periodic scrubbing, so media errors can be detected and corrected right away, and individual failed drives can be replaced.

But the winner for lotta-lotta-bytes storage is really still tape (with climate-control), as far as I'm aware. And multiple copies, stored in different locations if data is really important.

@MissConstrue

FWIW, my compromise (I have no tape drive) is I have a 4-bay NAS (proprietary , Synology) that does do regular data-scrubbing across its array.

I back up all my individual endpoint stuff to that NAS, as well as have some other things stored only on it.

Then I back up the NAS itself to set of single bigger HDDs that I rotate offsite, less often than I probably should .

@milomb @MissConstrue we use tapes a lot at work for cold storage (backup data with option to recover but rare access). It works very well, but takes hours/days to get the data back.
Do you know any reliable tape storage working as a service?

@lpryszcz
I don’t, sorry. I think Iron Mountain was the big one back in the day*, don’t know if that’s still the case.

* in the USA.

@MissConstrue Basically all forms of archival media need to be occasionally recopied, to last - a fact religious organizations have known for thousands of years already. It's just a question of what your timelines are.

Magnetic *tape* is not bad. With magnetic disk that isn't powered on, I'd be more concerned about catastrophic failure of the entire drive (turn it on after 20 years off, maybe it doesn't spin up) than "bit rot." RAID can help with that, but better if powered on.

@MissConstrue *Pressed* CDs and DVDs are critically different from *recorded* CDs and DVDs. The pressed ones are extremely curable, but can't be manufactured with one-off content.

And any kind of media, if keeping the data intact is critical, can be improved by recording redundant data (forward error correction) at creation time.

@MissConstrue I would suggest like others have to run a small system on the side that's just for file storage
You can get a second hand pc able to network and a couple hdds relatively cheap to keep them booted and accessible
You can then even run your own file on demand server or FTP so you can access the files online as well if you want.
It's not that expensive to run either.
HMU if you want to chat about it
@snek_mom And I have literally 10 or more old systems in my closet that can be cannabalized to create a server, now that you mention it.
@MissConstrue if you're looking at terabytes, I'd suggest a NAS, and non-SMR hard drives are the cheapest way. At least three bays lets you upgrade incrementally. Schedule tests - short SMART, long SMART, and file system scrubs.