On 5/7/2015 6:57 AM, Dave Cole wrote: > You asked about SSD drives as replacement for rotating disks. > > I've been using these: > http://www.amazon.com/gp/product/B00A35X6GM/ref=oh_aui_detailpage_o04_s00?ie=UTF8&psc=1 > > I've had some installed now for several years on commercial/industrial > machine that run daily and have yet to have a failure. > They are much more reliable than rotating disks which tended to sign off > after about 3 years.
Six SSD drives written to death. http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead It took a lot of data writing. For two of the drives more than two petabytes. Most people will never come anywhere close to using up the write lifetime of an SSD, but how much use one has had is a consideration if you're buying a used one. A common thing among the tested drives is their ends happened very rapidly once errors began to happen. But the worst thing about many SSDs is they brick themselves, either after a hard coded amount of data has been written or after a certain but unknown number of errors. The bricking happens on the next power cycle after the conditions are met. So you definitely want software monitoring the SMART status of SSDs because should the drive's condition deteriorate to self-destruct, you have one chance to save your date. Reboot and it dies. I assume that this self-destruct is for security reasons, to prevent people digging dead SSDs out of trash bins to see if there's any valuable data. You cannot recover data from a self-bricked SSD. Intel's commercial/industrial market SSDs don't self-brick, they just slow down write speeds to a crawl while still allowing the data to be read. They cost a lot and require a proprietary interface card. Intel also has revived the concept of the "hard card" with their high end SSD mounted directly onto a PCIe card. The intended market for it is in physically small servers but the "prosumer" and very high end personal computer builders have also been buying them. Doesn't matter to me if I wouldn't wear out an SSD, I wouldn't have one that bricks itself when it starts going bad, especially not one that bases the kill switch merely on how many writes have been done rather than accumulated errors. In the event of write wear or other damage, I'd want to be able to set the drive to read only so I can save the data then have a *user initiated* destruct that writes 0 to all blocks *then* bricks it. Would you want to have a car with an engine control computer than fried itself at exactly 300,000 miles or at the next time the engine was turned off after that distance? Most people will never drive a single vehicle that far, so why care about a limit? Another thing to consider is power off data retention. An SSD may not be suitable for a system that will be left powered off and unused for long periods of time or for offline data backup. I'd stay with magnetic storage for those uses. I've had very few (IIRC 6 or less in over 30 years) hard drives die in my computers. Most of them were secondhand, but two were brand new 15 gig IBM "Deathstar" models. One just went bad while the other would pass every diagnostic test and zero-fill and random-walk test I threw at it - but install an operating system (if it'd last long enough to get it installed) and it'd conk out in a few minutes. If I kept a fan blowing air on it from outside the case it would work as a data drive but not for Windows. I finally gave it away with a warning about how to treat it and to never put anything on it the new owner/victim cared about losing. The DeskStar debacle is why IBM sold off their storage division to Hitachi. Rather than fix the defects and replace all the bad drives with new models that were reliable, IBM decided to not even try to rebuild their reputation as a manufacturer of rock solid hard drives, just dumped ($2.05 billion is a big dump) it to end their losses. --- This email has been checked for viruses by Avast antivirus software. http://www.avast.com ------------------------------------------------------------------------------ One dashboard for servers and applications across Physical-Virtual-Cloud Widest out-of-the-box monitoring support with 50+ applications Performance metrics, stats and reports that give you Actionable Insights Deep dive visibility with transaction tracing using APM Insight. http://ad.doubleclick.net/ddm/clk/290420510;117567292;y _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
