[geeks] Disks: recommendations?

[Joshua Snyder]

On 10/30/2020 11:02 AM, Jonathan Patschke wrote:
> The seemingly-accelerated wear rate of moderns SSDs is exaggerated by how
> fast they are.B  In active use, an SSD under an extreme write-heavy load
> will fail faster than a mechanical disk in terms of calendar time, but
> generally not in terms of blocks written.

I remember seeing a Enterprise 15TB SSD that it's big claim to fame was 
it didn't have any write endurance "limitations". It only had a SATA 
interface so it was limited to ~600MB/sec interface speeds. Given the 
large size + the SATA interface and the 5 year warranty it was 
impossible to burn-up the drive while it was under warranty. So... 
marketing gave it "unlimited write endurance"!

> I think I've seen three device failures out of several hundred in two
> years.B  SSDs (from Micron/Crucial, Intel, WDC/SanDisk, and Samsung,
> anyway) are astoundingly durable.B  Remember to issue discard/dsm (a.k.a.
> "trim") commands for the unused portions of the media, and they'll last
> ages.
> 
> There are lots of low-end players in the SSD space (Adata, PNY, etc.), and
> my experience with those drives is that they are about as reliable as
> cheap thumbdrives.B  Do not trust them for data you care about.

I would agree that quality SSD's tend to be very durable. Early on there 
were some low-end players that I know had issues. I know people that had 
total drive failures with OCZ SSD's from the 2012'ish time frame.

Getting back to the original question. I have an Intel X-25 160GB SSD I 
bought in 2009. It's still going strong (73% life left currently). At 
one point this drive sat unused for about 3 years. All the data on the 
drive was totally fine.

Most of the wear on this drive came when I used it as a Zil and L2Arc 
for a ZFS file system.

> 
>> Do SSDs fail similarly, or do they just cross a line and go from
>> "working fine" to "completely dead" when their firmware decides it's had
>> enough?
> 
> SSD failure modes are manufacturer-dependent, but, generally speaking, the
> failures are either media, controller, or DRAM.B  Media failure manifest in
> stuck bits or a whole lost page (64MB or so at a time).B  DRAM failures
> manifest as unreliable transports.B  Controller failures manifest as a dead
> drive.
> 
>> I would hope they'd instead flip from "working fine" to "read-only", but
>> I have little faith such hopes would be realized.

I have seen both failure modes, drive is suddenly read-only and with the 
OCZ drives I talked about, they just stopped working entirely. They 
didn't even show up as drives when connected to systems.

			Josh