On 9 February 2011 02:57, Jacob Ritorto <jacob.rito...@gmail.com> wrote:
> Good stuff fellows; thank you. How bad are the failures resulting > from non-ecc and missing cache flushes? Can they put the pool into an > unrecoverable state, or is it just a risk of dropping a minute's worth of > writes or something like that? > > Again, this is a home archiving system with requirements of 1) data > integrity, 2) minimal impact on electric bill over a year's run time and 3) > approximately no noise. Into an unrecoverable state. That said, it is sometimes possible to get some data out of the pool, but it isn't a normal procedure (ie would require expert help). The way you lose only five seconds data is just when the atomic write didn't finish (ie power was pulled mid-write), which is how it's meant to be. More modern OSol/SX11 implementations have a default TXG sync time of 5 seconds, so this is the longest that the system can go between cache flushes (ie the most data you can lose). As for the noise, the pair of 3.5" 2TB drives will make more than any given low power pico/mini-itx board you find. Power: I'd suggest you work out what 25 watts over a year will actually cost you (~219kWh * your power cost). I suspect that paying more for a lower power (and hence computationally weaker) system is highly unlikely to pay for itself in any normal period. Non-ECC ram isn't a huge issue, until the RAM starts failing and throwing bad bits :P _______________________________________________ OpenIndiana-discuss mailing list OpenIndiana-discuss@openindiana.org http://openindiana.org/mailman/listinfo/openindiana-discuss
