2 ms·

Storage professional here. Disk drives failure rate is usually 0.5 to 3% a year. Drives do fail, so you can't draw any statistical significant figures from the

by wazoox 15y ago

Storage professional here. Disk drives failure rate is usually 0.5 to 3% a year. Drives do fail, so you can't draw any statistical significant figures from the failure of a couple of drives among a total of four or five (or even ten or twenty). However my company's selling drives by the thousands, and here's what I can say: Seagate barraccuda are the worst pieces of crap being sold since the infamous IBM 80GB debacle; their failure rate is 10% a year, no less. Unsurprisingly Seagate just dropped their warranty to ... 1 year! Keep away from these at all costs... Seagate Constellation seem to be OK but I only have a few of these. Hitachi 1, 2 and 3 TB look impressive: 1 failure among 2 thousand drives, in 2 years. WD seem OK, but I haven't nearly enough of them to get significant numbers (a couple hundreds).

ugh 15y ago
Cool! Do you have any numbers for SSDs?
- wazoox 15y ago
  Unfortunately I didn't set up as many of these, but the first generation of SLC SSD were quite prone to failure; The latest SSD with Sanforce 12xx controllers seem to work very well. I had no failure on these yet, though I had some trouble with older designs.
pasbesoin 15y ago
Unfortunately, I recently picked up a couple of Barracuda because of a price break fluke that let me essentially escape the flooding price inflation. I guess I will use them with trepidation (redundant backups, mostly sitting on a shelf) and plan to phase them out in the coming year and/or into transient cache. Thanks for the heads up. P.S. I was set to go with Hitachi (2 TB 7200 RPM) when their prices more than doubled on me. Price aside, are they still a good bet like they were a couple of years ago? (I recall reading that at one point Google was favoring them.) Also, since I'm adding a postscript, does anyone recall the HD failure rate study that someone put out a few years ago? Of course, things change, but behind the particular models in seem to recall some useful inferences about company/brand quality/reliability.
- wazoox 15y ago
  There were several interesting studies these past years : one from google, one from NetApp, an older one from the CERN: Google 2007: http://research.google.com/pubs/pub32774.html http://research.google.com/pubs/pub32774.html Carnegie Mellon 2005: http://www.usenix.org/events/fast07/tech/schroeder.html http://www.usenix.org/events/fast07/tech/schroeder.html CERN on data integrity: http://indico.cern.ch/getFile.py/access?contribId=3&sessionId=0&resId=1&materialId=paper&confId=13797 http://indico.cern.ch/getFile.py/access?contribId=3&sess... NetApp 2010: http://synergy-ds.com/netapp/wp-7005.pdf http://synergy-ds.com/netapp/wp-7005.pdf
  - pasbesoin 15y ago
    I may be remembering the Google study and/or something else from about that time frame. I have a vague recollection now of both Google's opinion and and something else that had at least some degree of separation from Google. Thanks very much for the informative response. I'll have a look at these (as soon as I talk myself down sufficiently from my guilt over my price-motivated Seagate purchase ;-). In part, I also pushed the button on Seagate (despite the 1 TB "click of death" episode from a few years ago) because a recent WD laptop HD purchase went rather bad with incessant parking that is apparently inherent to the model's firmware. I finally fixed that on the XP Pro machine it lives in by finding and setting to run on start-up: http://sites.google.com/site/quiethdd/ http://sites.google.com/site/quiethdd/ [1] But not before SMART reported significant aging for the drive. Which rather pissed me off. (I should pull it and RMA it -- 5 year warranty.) So, that helped lead me to Hitachi. Then the price inflation, plus the news that WD is buying Hitachi, and I figured I'd give Seagate another go. I hate purchasing consumer electronics, these days. -- 1. Which I found at about the same time I learned I could tweak the relevant settings manually -- so, I went with the pre-packaged implementation.
listic 15y ago
Which Seagate Barracuda models exactly are bad, in your experience? I think we cannot generalize across all models and generations, can we?
- wazoox 15y ago
  The worst are the 1TB ES2. These are the same as 7200.11 in the consumer series.

by wazoox 15y ago

ugh 15y ago

Cool! Do you have any numbers for SSDs?

wazoox 15y ago
Unfortunately I didn't set up as many of these, but the first generation of SLC SSD were quite prone to failure; The latest SSD with Sanforce 12xx controllers seem to work very well. I had no failure on these yet, though I had some trouble with older designs.

pasbesoin 15y ago

Unfortunately, I recently picked up a couple of Barracuda because of a price break fluke that let me essentially escape the flooding price inflation. I guess I will use them with trepidation (redundant backups, mostly sitting on a shelf) and plan to phase them out in the coming year and/or into transient cache. Thanks for the heads up. P.S. I was set to go with Hitachi (2 TB 7200 RPM) when their prices more than doubled on me. Price aside, are they still a good bet like they were a couple of years ago? (I recall reading that at one point Google was favoring them.) Also, since I'm adding a postscript, does anyone recall the HD failure rate study that someone put out a few years ago? Of course, things change, but behind the particular models in seem to recall some useful inferences about company/brand quality/reliability.

wazoox 15y ago
There were several interesting studies these past years : one from google, one from NetApp, an older one from the CERN: Google 2007: http://research.google.com/pubs/pub32774.html http://research.google.com/pubs/pub32774.html Carnegie Mellon 2005: http://www.usenix.org/events/fast07/tech/schroeder.html http://www.usenix.org/events/fast07/tech/schroeder.html CERN on data integrity: http://indico.cern.ch/getFile.py/access?contribId=3&sessionId=0&resId=1&materialId=paper&confId=13797 http://indico.cern.ch/getFile.py/access?contribId=3&sess... NetApp 2010: http://synergy-ds.com/netapp/wp-7005.pdf http://synergy-ds.com/netapp/wp-7005.pdf
- pasbesoin 15y ago
  I may be remembering the Google study and/or something else from about that time frame. I have a vague recollection now of both Google's opinion and and something else that had at least some degree of separation from Google. Thanks very much for the informative response. I'll have a look at these (as soon as I talk myself down sufficiently from my guilt over my price-motivated Seagate purchase ;-). In part, I also pushed the button on Seagate (despite the 1 TB "click of death" episode from a few years ago) because a recent WD laptop HD purchase went rather bad with incessant parking that is apparently inherent to the model's firmware. I finally fixed that on the XP Pro machine it lives in by finding and setting to run on start-up: http://sites.google.com/site/quiethdd/ http://sites.google.com/site/quiethdd/ [1] But not before SMART reported significant aging for the drive. Which rather pissed me off. (I should pull it and RMA it -- 5 year warranty.) So, that helped lead me to Hitachi. Then the price inflation, plus the news that WD is buying Hitachi, and I figured I'd give Seagate another go. I hate purchasing consumer electronics, these days. -- 1. Which I found at about the same time I learned I could tweak the relevant settings manually -- so, I went with the pre-packaged implementation.

listic 15y ago

Which Seagate Barracuda models exactly are bad, in your experience? I think we cannot generalize across all models and generations, can we?

wazoox 15y ago
The worst are the 1TB ES2. These are the same as 7200.11 in the consumer series.