On Mon, Feb 13, 2017 at 4:49 PM, Mick <michaelkintz...@gmail.com> wrote:
> On Monday 13 Feb 2017 13:17:14 Poison BL. wrote: > > On Mon, Feb 13, 2017 at 11:44 AM, Daniel Frey <djqf...@gmail.com> wrote: > > > On 02/12/2017 02:40 PM, Alan Grimes wrote: > > > > So does anyone have any evidence of a current generation SSD lasting > > > > more than 20 days? > > > > > > I have tried various SSDs (multiple brands and generations) over the > > > last maybe five years and found that they're very unreliable (multiple > > > brands too.) > > > > > > I know everyone's saying these things are reliable but out of four SSDs > > > I own, I've had to replace three, some more than once. > > > > > > I just don't use them for anything I want to stay working. Right now I > > > keep them in my mythtv frontends as I can restore the OS easily. One > one > > > of them the company involved (Kingston) even sent me a newer > drive/model > > > as it was replaced more than once. > > > > > > I know they're fast. But what's the point of going 500 MPH and crashing > > > into a mountain with no chance of repair/recovery. I went back to a > > > (relatively) slower rust raid10, and it's been reliable for the last > > > four years. At least with a hard drive failure, you stand /some/ chance > > > at recovery, not zero. > > > > > > The one SSD that hasn't had to have been replaced under warranty is in > > > my laptop which I generally use maybe a dozen times a year. I fully > > > expect it to die one of these times when I boot the laptop (it's one of > > > the old models.) > > > > > > My experiences are with Samsung, Kingston, Intel, Crucial and AData > > > SSDs. The last one I bought because these things I view as throwaway > > > devices (the warranty expired on the original Crucial) and don't want > to > > > spend big money on them. I have noticed the AData SSD's performance is > > > not as fast now as it was new (maybe 1.5 years ago?) So it'll probably > > > pack it in soon too. > > > > > > Dan > > > > I've had more than one spinning rust drive fail hard over the years as > > well, though yes, you do usually have some chance of recovery from those. > > Gambling on that chance by leaving a given disk as a single point of > > failure is still a bad idea, spinning disk or not. The point that you > went > > from single-disk SSD back to raid10 makes me question why, if your uptime > > requirements (even if only for your own desires on a personal machine) > > justify raid10, you weren't on at least raid1 with the SSD setup. > > > > As for performance degredation on SSDs, that I've definitely seen on > pretty > > much every brand, though I've had good luck doing clean reloads on > samsungs > > once or twice to get speeds back up some (somehow, even trim doesn't seem > > to keep things at their best). > > > > I can't say they're more or less reliable than spinning disks, though > they > > do have the benefit of no moving parts to wear out over time (thermal > > cycles can still cause a physical failure on them, though). > > Have you noticed a difference between mounting partitions on them with the > discard option, Vs running fstrim on a cron job? > -- > Regards, > Mick I actually only have one (exceptionally cheap, and little used) in a linux box at all, and haven't tested with anything other than having discard set. I sadly have to live the windows life on all my work machines :( -- Poison [BLX] Joshua M. Murphy
