I sure hope that the figures quoted in the below video are lower than actual life expectancy. I do not relish the thought of most of my farm’s storage hardware dying in two or three more years.
And then there is data fade, for which the remedy is to re-write the data. That is simple enough, for a few drives. But when you have dozens of drives, or hundreds of drives, then what?
The video’s host recommends re-writing the drive’s data every five years.
Anyone going to do that? I never heard of anyone actually doing that.
Normal lifespan of a drive is a few years if you’re unlucky in my experience. When making infrequent reads for Chia I would be shocked if they start dropping like flies in that time span. The workload is extremely light.
Good one…so far I rewritten data on drives several times over a few yrs of Chia…not of my choice really, but needed to keep up with so-called ‘advances’. So as Ronski says…OG plots > NFT plots > BB plots > Cuda plots. At this rate data will never get a chance to ‘fade’ lol! And that’s just for Chia blockchain and what I remember as it all gets to be a bit of a blur … all the replotting Plus I doubt the replotting will stop…uncompressed plots soon…mark my words.
Unless we are expecting to re-plot our farms every few years, then the fading can be a serious issue, if what the host of that video said has merit.
I have had enough of re-plotting. Unless it is going to bring in substantially more income, it is not worth the effort (speaking for myself).
I have never heard of anyone losing data from aging drives that have dormant data, due to magnetic fading.
Drives do fail. But is magnetic data fade really a problem that can rear its head in 5 or so years?
Between drives having a ~5 year life expectancy, and now data fading, what’s next? Drives exploding?
I am counting on my drives lasting long beyond 5 years, with no magnetic fading causing any problems. But that is from my own personal experience with drives, and wishful thinking. That video might have been overly pessimistic. The host seems to be knowledgeable, but maybe he missed the mark with that video?
And since no one has chimed in with any reports on losing data due to fading, I think that it is not a serious concern.
Magnetic fading causing bit rot is not a HD failure (unless it is a catastrophic failure of a platter what will most likely happen right away; I don’t think the U shape failure rate applies to magnetic fading, or rather is as much pronounced). In the best-case scenario, either FW or file system will recover lost bits (if ECC is in place). It may require disk scrubbing / testing, though. In case such cell will be recovered, one may expect a bit rot in the same cell roughly after the same amount of time (as such cell has faster magnetic fading). In case a cell cannot be revived (cell doesn’t hold the value anymore), the cluster will be swapped for a spare one ending with a damaged file. Once the cluster is swapped, the HD condition will be as good as before. One very rough estimation how many weak cells may be on a disk is the ratio of spare clusters to the total size of HD (a tiny fraction). Also, magnetic fading is a property of magnetic storage; however, it usually happens after a few decades (several times longer than the MTBF).
There are no similarities between “premature” magnetic fading and SSD cell charge leakage. The first one is due to platter manufacturing (inconsistent magnetic layer), where the second one is an accumulated damage on every single cell write.
If fading was a thing, it would be widely talked about and a known fact. Hard drive technology is how old? Let’s be real. Of course SSD data does deteriorate over time. See? We all know about that because it’s a thing. Hard drive technology has only gotten better. To think there is some spooky flaw that’s going to bite every hard drive owner in the a$$? I haven’t watched the video yet. I still stand by my points.
Plus I doubt the replotting will stop…uncompressed plots soon…mark my words.