I think the only way to get an unbiased answer is to somehow compare MTBF and AFR between HDD and SDD. You then take into account environmental conditions and see which best suit. For example, a Laptop is more likely to get a shock, so a SSD is better than a HDD as it is more resistant to shock. Shock is not a problem for desktops, the possible problem there is heat. HDDs don't like getting too hot, so if you don't wish to have good air flow/noisy fans etc, and SSD might be the better choice. SSD seem generally better in adverse environmental conditions. I have seen proof that things with moving parts are more likely to fail. Anyone who has deployed thousands to fanless thin clients with no HDD will know that the failure rate is far less than desktops with fans and HDDs. So, my gut feeling is SSDs should be more reliable than HDDs. What I have not seen is SMART stats from SSDs telling us about SSD specific items. E.g. Some stats or values for an SSD for how many failed storage spots there are. I have not seen the Linux smartctl program list anything like this for SSDs, but I have not checked the SMART standard, to see if stats for SSDs are reported or not.
What is definitely don't like with reports of SSDs up till now, is that they quite often go dark without any notice. I.e. You loose everything with no warning. That is a failure mode that is very unfortunate. I think that if manufactures could do something to mitigate the "going dark" failure mode, then at least when the SSD fails, you could still recover data from it, much like you can do with most failure modes of a HDD. I saw this with some USB flash sticks. Their failure more is they just switch to read only. Another one I saw, for a military app, was a test point on the SSD, so you could access the flash chips directly without having to de-solder them. It made data recovery in case of a fault a lot quicker and could be done by non-experts. It would let you read the entire flash contents, bit-by-bit. You then ran it through a program that contained the rules for the data layout, and the data could be recovered easily. You could not do this with HDDs, because you need experts to deal with the platters when recovering data from HDDs. I would really like to see that feature on consumer SSDs. It completely removes the "going dark" failure mode. Kind Regards James On 27 September 2013 13:55, Gordon Scott <gor...@gscott.co.uk> wrote: > On Fri, 2013-09-27 at 13:41 +0100, Alan Pope wrote: > > > > > I've seen reports of SSDs failing too. I currently have 7 from various > > manufacturers in place in different machines. Not had a single one > > fail yet. I've had my fair share of rust go bad. > > The people with the problems tend to make the most noise. > Of course. > > > I realise my anecdotal evidence is meaningless, but so long as you > > have good backups a failed SSD is about the same inconvenience as a > > failed hard disk. > > Well that's good to hear, and as it should be. > > BTW, I never meant to imply that SSDs were bad in any way, just that > they're not necessarily as good as one might expect of solid-state. > And they're still quite a bit more costly than rust, byte-for-byte. > > Gordon. > > > > -- > Please post to: Hampshire@mailman.lug.org.uk > Web Interface: https://mailman.lug.org.uk/mailman/listinfo/hampshire > LUG URL: http://www.hantslug.org.uk > -------------------------------------------------------------- >
-- Please post to: Hampshire@mailman.lug.org.uk Web Interface: https://mailman.lug.org.uk/mailman/listinfo/hampshire LUG URL: http://www.hantslug.org.uk --------------------------------------------------------------
