Re: RAID question - dying
On Fri, 05 Dec 2003 17:52:19 -0800, Alvin Oga wrote:
> On Fri, 5 Dec 2003, Paul Morgan wrote:
>> > - disks doesn't fail as often as a power supply
>> > -- hint.. add a nice fan for each disk
>> My experience is the opposite: I have never had a power supply fail, but I
>> have had several disks fail.
> everybody's experiences will be different because:
> - some people buy good products, some people buy cheap stuff
> - some people add fans where needed, some people dont
> - there are always exception to the rule of "good stuff"
> ( like ibm's major !@#$% with their deskstar series of drives,
> ( caused them to sell their entire disk drive division to hitachi
> ( -- just kidding .. but the timing is good for a joke
> given the same system ... things that should die first would be based
> on MTBF ... ( mean time between failure ) all rated at 25C
> Warranty period is probably closely related to MTBF ...
> - fans ( typically no warranty ... )
> - power supplies ( some have 100,000 hrs if you believe it )
> which confirms Pauls experiences w/ power supply vs disks
> example: http://www.sparklepower.com/pdf/FSP300-60GRE.pdf
> - disks ( typically warrantied for 1-3 yrs )
> - cpu has an mtbf around 30,000 hrs ( about 5 yrs )
> -- notice that your cpu lifespan dies 2x faster for each
> 10C increase in temp
> c ya
Since 1982 when I bought Chuck Peddle's Sirius 1 (Victor 9000 in the US),
I have had the following failures:
power supply: 1
hard disk drive: 4
modem: 1 (lightning)
I've been building my own machines for a number of years, and have
generally had more disks than power supplies.
At the moment, for instance, I am running 1 power supply, 1 CPU and 6
disks, so my odds in favor of a disk failure as opposed to a power supply
failure are significantly increased.
Your response was interesting, alvin, thanks.
"The number of UNIX installations has grown to 10, with more expected."
(The UNIX Programmer's Manual, 2nd Edition, June 1972)