Leave it connected and it's vulnerable to wearing out, lightning, malware and system failures. I did recover a HDD once where PSU failed and bits physically blew of most chips. I had a same model with too many errors and working PCB, but that won't work with all drives.
An off-line HDD can easily last 50 years. But the issue is that the interface can become unavailable. SATA-IDE adaptors seem to work, SCSI cards still exist, though cable adaptors may be required. But I've had HDD that look like IDE and are not. Also good luck trying to read IBM XT HDD (MFM dual ribbons). So from time to time you need newer media with newer interfaces while you can still read the old media.
Quote:
|
But more generally I'd avoid brand new drive models (which usually means avoiding the latest ultra-high-capacity stuff anyway),
|
That's what I meant by high density.
You can't keep helium in or out. The 200 years is probably a lie. Old CRTs even get poisoned by helium because with vacuum inside and 1 atmosphere outside it alone of all gasses seeps through the glass. You can't seal in helium unless it's at a very low pressure, but then the heads can't fly. They use Bernoulli effect. It's akin to Concorde at hedge height. Hence the cleanest places are not chip processing (but awesomely clean) but HDD assembly.
Quote:
|
Never tried that, but then if I *do* have time-delayed ransomware I think I'd notice it when it detonated for the first time and would intentionally restore (off the offsite disk) from an older backup
|
Well, how time delayed is the Ransomware? The clever stuff encrypts the backup and decrypts it when you restore on the same machine, till it goes off.
You can't have a true disaster recovery plan without at least one spare PC offsite anyway. Unlike AV such a setup will 100% detect ransomware running that's not "detonated".