On 15/09/2014 13:10, Stroller wrote: > > On Sun, 14 September 2014, at 9:53 pm, Alan McKinnon > <[email protected]> wrote: >> ... >> Google has 1,000,000+ drives, I'll trust what they say after statistical >> analysis. >> Rack Space has a goodly number of drives too so I'll trust them as well. >> I'll even trust my previous employer (an ISP with 10+ data centres) and >> customers fitting every example of every drive out there at random. > > There is some great information of this kind available - the trouble is that, > by the time you've tested drives for 3 years, your information is 3 years out > of date (as far as "what's the latest drive I should buy?" is concerned). > > http://www.pcworld.com/article/2089464/three-year-27-000-drive-study-reveals-the-most-reliable-hard-drive-makers.html > > From this report we should buy Hitachi and distrust Seagate, but not only > were only specific models tested, for all know both manufacturers may have > long ago changed their manufacturing methods now.
Which is why most folks who buy substantial numbers of drives do something like this: 1. Decide what drives[1] you like and take proper statistical history into account. The answer is often somewhat random and more about "I like" rather than "I know for a fact". 2. Buy those drives. 3. Establish a relationship with that vendor. 4. If step #2 goes south and you got duds, get replacements leveraging on #3 But asking a random bunch of dudes on a mailing list "what is a good drive right now" is a useless question. If the mailing list is big enough there are only two eventual answers: - any of them - none of them [1] This can be 0, 1 or more drive types -- Alan McKinnon [email protected]
