New SSDs from crucial
- Thread starter andrewbedia
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
I don't think I've heard the whole story about this reliability problem everyone seems to keep talking about on STH. Everywhere else says they are excellent.
I'm all ears. I don't even own one. I own three OCZ drive, an intel drive, a samsung drive, and a kingston drive. I don't really have a vested interest in disagreeing with you, but I would like to hear STH's side of the story here.
I'm all ears. I don't even own one. I own three OCZ drive, an intel drive, a samsung drive, and a kingston drive. I don't really have a vested interest in disagreeing with you, but I would like to hear STH's side of the story here.
I have not had issues with reliability and I have quite a few of the Crucial drives. What I have seen is that the M500 480GB drives (overprovisioned down to 420GB) still have some long service times on read/ write requests. The new servers are in the DC which will bring higher-end SAS/ SATA (S3700) drives to fix the service time problem which is impacting a non-negligible portion of traffic to the sites.
Well, for me, I've had 100% failure rate on the MX100. Sample size of 1x 480GB unit. Windows locked up, upon reboot the drive was missing. Took a cold reboot to come back to life. Did that randomly for a week then finally died. Tested on several known good machines.
Replaced the drive with a Sandisk UltraII and it's been perfect so far.
That's my anecdote.
Replaced the drive with a Sandisk UltraII and it's been perfect so far.
That's my anecdote.
This - and doubly so on SAS controllers. I've found the M500's to be one of the most stable drives. At work there's maybe 200 and we've had 1 rma.
I'm still buying Samsung or Intel drives.
You must not do write workloads on them... Might I ask what servers and what controllers?
I have seen signal integrity issues (negotiate at lower link rate) and iop hangs...
Sample size 20 or so drives to track down said performance anomalies.
Wow, that's really impressive. Like @Patriot, I'm curious what kind of workload you're doing on them.
Customer specified... can't go into details but they were FIO sync lib .... so Q=1 ~3k iops normally but GC would fire up and drop it below 10...
Not really a heavy write workload and they still crapped out. Continuous yes... but not high rate.