Andrej Ricnik-Bay wrote: > In my testing I found that once you hit 10 spindles in a RAID5 the > differences between it and a RAID10 started to become negligible > (around 6% slower on writes average with 10 runs of bonnie++ on > 10 spindles) while the read speed (if you're doing similar amounts > of reads & writes it's a fair criterion) were in about the 10% region > faster. With 24 spindles I couldn't see any difference at all. Those > were 73GB 15K SCAs, btw, and the SAN connected via 2GB fibre. Isn't a 10 or 24 spindle RAID 5 array awfully likely to encounter a double disk failure (such as during the load imposed by rebuild onto a spare) ? I guess if you have good backups - as you must - it's not that big a deal, but I'd be pretty nervous with anything less than RAID 6 or RAID 10 . -- Craig Ringer -- Sent via pgsql-performance mailing list (pgsql-performance@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-performance
