home | list info | list archive | date index | thread index

Re: [OCLUG-Tech] SSD wear

Hi Rick,

> On Jan 13, 2016, at 14:36, Rick Leir <rleir [ at ] leirtech [ dot ] com> wrote:
> Hi Spencer
> Yes, I heard that you get no warning, no smartd error counts or whatever. Suddenly there is no response from your disk?

Yes. Unlike spinning rust which can grind (literally sometimes) for a bit longer and if you stick it in a freezer, that might extend it’s life long enough to get your data off, my SSDs failed within a few minutes. It’s been a few years so I forget the details. I’ve had both HDD and SSD fail while traveling so I always travel with a complete backup now.

SSDs compared to a HDD is like an endorphin rush, when all the sudden, your so-so 2 year old computer kicks ass. I stuck a older SSD into my rather average Dell desktop. It turned into a box which has a lot of “snap”.

One side note, just because I managed to pull data off a dying HDD was actually very dangerous. There was enough corrupted data that I am still cautious about the data which I recovered due to possible corruption. Only btrfs and ZFS do any data integrity checking to my knowledge and I wasn’t using either. My NAS now use ZFS.

Another caution about SSDs, they MUST be kept fairly cool or they will die of premature death. Heard via H/W friend about Micron admitting that. I try really hard to keep the mSATA SSDs under 50C in our systems ie. fans speed is keyed to SSD temperature, not CPU temperature.

I’ve had mSATA SSDs that runs slower and slower as the internal temperature increases and ended up with lots of errors by the time it got to 80C. Some form of cooling is required for SSDs. 2.5” SSDs are better than mSATA because they have more thermal mass. Samsung M.2 SSDs runs cooler than Micron mSATA SSDs.