And would like to replace it. What would you recommend in the 1/2 to 1 Tb range for long term durability? An actual spinning HD or an SSD?
Most modern SSDs will be multilevel cells (3 or 4 bits per cell). Since flash loses charge over time, having to detect multiple charge levels in a cell implies that it's more likely that you'll encounter bit flips (vs SLC flash). I don't see how this is addressed in powered devices because charge loss will still occur and won't be "addressed" until a block is rewritten. I've seen guaranteed data retention time for unpowered devices as little as a year (it depends on P/E cycles, bits per cell, storage temperature, etc.).But my question was more about the reliability of a HDD vs a SSD?
SSD, they have surpassed HDDs some time ago, overall, your more likely to get a failure with an HDD than SSD. But what are you backing up? why not just use the cloud (or Github for code)?That was my exact thought, and the reason I deliberately chose a storage capacity under 1Tb.
But my question was more about the reliability of a HDD vs a SSD?
All digital storage is unreliable. It can, and does, fail at any time.That was my exact thought, and the reason I deliberately chose a storage capacity under 1Tb.
But my question was more about the reliability of a HDD vs a SSD?
In terms of performance, yes. In terms of reliability, I think not. A hard drive used for long term backup won't be spinning constantly, so could retain data for many decades. I don't think the same could be said about flash.SSD, they have surpassed HDDs some time ago, overall, your more likely to get a failure with an HDD than SSD.
The cloud doesn't have as much privacy and security as a local storage medium. About the only reason I'd use the cloud is for sharing data.why not just use the cloud
You might be right, he did state "backup". My PCs have SSDs and they have not suffered in any way for many years.In terms of performance, yes. In terms of reliability, I think not. A hard drive used for long term backup won't be spinning constantly, so could retain data for many decades. I don't think the same could be said about flash.
The cloud doesn't have as much privacy and security as a local storage medium. About the only reason I'd use the cloud is for sharing data.
Once something is stored in the cloud, you've lost control over how long data is retained and who can access it. Server farms have their own backup and retention methodologies. When I was a systems manager, I stored one monthly backup per year for many years. If you deleted something from cloud storage, you have no idea how many years it will still be available on backups.
When I was a systems manager (early to mid 1980's), I stored one copy on-site and one off-site at a company that specialized in storing tapes. I was in the SF Bay Area and the off-site storage was around Sacramento.Back in the day we'd have a firm come over every night (this was the City of London) and collect backup tapes of vital financial positions. I guess these guys had some safe location for them - this was mid 80s.
Could you explain?The fact that, after about eleven years, I still have 3.5 GB of space available on it, despite having the final back ups from four other computers as well as my phone, makes me opt for reliability over capacity (higher capacity virtually always means that some reliability is being sacrificed).
To get significantly higher capacities, you use smaller geometries or more aggressive multi-level storage, both of which reduce noise margins.Could you explain?
My old raid media backup device finally died but the computer originally using it died long before.Every computer I have ever owned has outlasted all the devices I bought on which to back it up.
SSDs utilizing multiple bits per cell all use error detection. When a block is determined to have too many errors, the block is marked bad and the data is rewritten to a block from the overprovision area. Presumably if the overprovision area is exhausted, a "regular" block would be used.Since flash loses charge over time, having to detect multiple charge levels in a cell implies that it's more likely that you'll encounter bit flips (vs SLC flash). I don't see how this is addressed in powered devices because charge loss will still occur and won't be "addressed" until a block is rewritten.
One SSD manufacturer recommended that unpowered SSDs be powered up (and somehow scanned/initialized) at least once a year to prevent data loss.I've seen guaranteed data retention time for unpowered devices as little as a year (it depends on P/E cycles, bits per cell, storage temperature, etc.).