Archiving HDD vs. SSD in 2020-2021?
Hey all!
I asked this question a few years ago and wanted to ask again because... technology changes. Years ago the recommendation from fine folks here was that HDDs are more reliable than SSDs for long term storage, or "archiving." Meaning, for me at least, the traditional 5-year window between data migrations. My questions are simple. Is that still the case? If so, any idea how much longer until SSDs are equal to HDDs in this area, if ever? How do we feel about SSDs replacing HDDs in this regard? I feel like there may be some old school technical reasons to never abandon HDDs (just seems like one of those things). Is that right or can I look forward to a switch someday, price notwithstanding? Thanks! |
Much of the dabate over using HDDs vs SSDs for long-term storage revolves around the nature of how each type of drive fails over time. If HDDs experience failures they often occur over a long period of time. Also if part of a hard disk drive isn't operating properly, it can likely still power on and data can be extracted from it. SSDs fail much more suddenly and the process of recovering data from dead SSDs is difficult/much more expensive.
|
Thanks! I had that general understanding. Do you have an opinion on improvements over time? I'd love to know how the gaps might be closed - improvement in circuit reliability; data retrieval costs coming down; improvements in alerting to imminent loss or corruption, and so on. Certainly prices for new SSDs coming down someday, to where they're comparable to HDDs, would make things easier for consumers to double up on redundant drives to compensate for sudden and complete losses (assuming no other improvements).
I mean, SSDs are the future. There must be some ways of improving their reliability. At least across that 5 year window. I'm also interested in opinions on general improvements, such as read/write, wear (durability, performance, etc). Things that have typically made them less desirable compared to HDDs. |
I've had dozens of HDDs over the years and only one failure (as mentioned, over a few days and I was able to have the data retrieved by an expert). I've had 8 SSDs in the last 3 years and two have died, instantly. Completely wiped in a second. I would never put anything critical for storage on an SSD.
|
Nothing has changed in at least 5 years now. Some larger drives, new methods to allow it. That's really it.
Same advice still applies. Multiple copies, on multiple media types. - use DVD-R (or +R), less so BD-R -- obviously smaller, so backup only most important, not everything - use hard drives (multiples) - never flash memory - SSD should is working space, not for archive/storage SSD have probably gotten worse in recent years, not better. Pushing more data results in slowdown and potential integrity compromises. QLC NAND is awful. I've not moved past using Samsung 850 EVO drives as a result (TLC). I don't understand why companies insist on 2.5" HDD SATA form factor when you could stick lots of TLC into a cube or thicker platter. Something larger in size, both physical and data size. NAND prices are falling, after all. |
Quote:
|
Quote:
|
Thanks LS. Another question about data migration: if I have HDDs in storage, unpowered and put in use simply to add new files to them when I go there (the storage unit is in another town so it's only a few times a year), is it recommended I still observe the five year window? Or can I stretch it out a couple more years?
For that matter, it IS five years still, correct? -- merged -- Bumpity bump bump? :) |
Stretch it longer. Just boot up several times per year, access, maybe even scan drive.
|
Thanks LS. And congrats on post #10K.
|
Quote:
|
Quote:
Quote:
|
There's no perfect solution to the ages old archival question.
Quality of consumer grade HDDs has dropped in the past five years resulting in far more catastrophic failures than I have ever seen in the years prior. It use to be true that if your mechanical HDD failed there was a high probability of being able to recover most of your data from it. Now, the kinds of failures we're baring witness to are utterly absurd things that should never happen, platter damage! Platter damage is absolute and nothing can be retrieved from a damaged platter. On the flip of that, SSD/memory chip data recovery technology has improved. It's still true that if a memory chip fails you lose the data that was on it and any overlapping files partially on that chip, but that's just the nature of how these things work sadly. Best advice still holds true, back-up your important data to more than one device. I do not recommended those stupid consumer model pocket external HDDs to anyone though, they're junk. Use either desktop 3.5" HDDs or 2.5" SSDs either internally or via a USB HDD Docking station, and have a safe place to store the drive(s) after performing your back-up. If you want to go the optical disc route, M-DISCs are the only real option. All other optical discs (DVD-R/+R/RAM and CD-R/RW) are extremely prone to a unique form of so-called "disc rot" from their chemical layers that the data is "burned" into degrading over time. In the worst of the worst I have had data discs suffer disc rot withing mere weeks of burning data onto them, while other discs have mysteriously managed to retain their integrity for over ten years now. Point is, it's very unpredictable how or when a writable disc will degrade. M-DISCS Are unique in that they were created for data preservation and are made differently, and also require a ODD that is marked as M-DISC ready as the writing laser uses a different intensity setting for writing data onto a M-DISC. M-DISCS Are not cheap however, when I was working at a electronics retailer just three years ago, the per disc cost averaged $5, but this may be a worthwhile price to the data you wish to preserve. And your third option is data cassette back-up. Not the cheapest alternative anymore with falling SSD/Flash drive prices. Specifically for entry level DAT72 and DAT160 drives are the best choice as they can be purchased with a USB 2.0 interface, whereas the rest of the currently available stuff is SAS and Super SCSI interfaced, the sort of things for pro-grade servers. DAT72 Uses 72GB Data Cassettes, and DAT 160 uses 160GB Data Cassettes. |
As I have done for the last 5 years, I used a hard drive stored in a metal container to avoid problems with magnetic fields. A safety deposit box is the perfect place for such drives.
The short life expectancy of data readers is a major problem with digital storage. For this reason, the Library of Congress does not consider any digital medium suitable for archival purposes. Physical negatives and physical prints will last (with the exception of inkjet prints which are more easily damaged). |
The more advanced the technology the more fragile it is.
Bury a clay tablet in the desert. Subject it to war, floods, cold, heat and break it in multiple fragments. You can still get the data off that clay tablet. Cut a VHS tape into pieces and give it environmental hell. With a little effort you can probably get the data from that tape. Scratch a DVD and you’re screwed. |
Quote:
Quote:
Quote:
Quote:
Their is no such thing as "disc rot" for CD/DVD/BD. That unofficial term (also "laser rot") was unique to the Laserdisc format, and involved oxidation and self-destruction of the foil layers. It had nothing to do with dye, dye-based media, or CD/DVD/BD. It cannot be extrapolated. M-Disc is not a recommended DVD. Poor reflectivity, closer to phase-change RW. Quote:
Quote:
Quote:
The same BS existed for gold media, which also had lower reflectivity than standard silvery metals. Quote:
Quote:
Quote:
|
Quote:
The firmware chip you referred to on the modern drives actually simplifies data recovery for a data recovery professional should their services be needed as a physically identical board from a different HDD can be flashed with the firmware required by the drive that data needs to be retrieved from. But going back to the catastrophic failure types being encountered in the modern drives, they're literally engineering and/or quality control failures. The R/W heads on the armature of a HDD should never separate from the armature and scrape the platter. The drive motor should never seize, overheat, and melt the platters. We should never open up a failed drive that in testing is mechanically sound but unable to read any data only to find aluminum flakes left over from the chasis machining process are still inside of the drive and have scratched the platters. Yes, finding DOA HDDs, and other components, has and will always be a issue, but I;d rather have DOA components that only create a temporary setback than poorly made functional equipment that suicide themselves and take a clients valuable data with them. Quote:
Quote:
Quote:
Quote:
Quote:
|
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Even a stone tablet isn't just a stone tablet. Limestone vs. clay. Quote:
Quote:
Quote:
Quote:
|
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.