i've been curious about relative performance of 4 drives (or sets of drives):
1) intel x25-m ssd
2) wd5000aaks 500GB single drive
3) 4 wd5000aaks 500GB drives in RAID0 formation
4) apple time capsule (2TB, connected via gigabit ethernet)
i know that xbench is not the last word on performance, but since what mattered here was relative performance, i thought it was ok. i am currently using 10.6.1 on a mac pro.
results for the disk component of xbench?
1) intel ssd - 257.08 [amazing in random access]
2) single 500gb WD - 77.63
3) RAID0 - 229.84 [consistent between sequential and random]
4) time capsule - 16.82 [random performance was better than sequential, and bigger blocks yielded higher performance than smaller ones]
would you expect a single drive connected via gigabit ethernet (time capsule, in this case) to score only about 20% of an internal drive -- and about 7% of that of an SSD?
i guess the question comes down to how much gigabit ethernet's overhead is a constraint in terms of disk performance. i wonder if there are any setting i could change, or if it could simply be a constraint of the router on apple's time capsule, or if this is the best a single network drive could be expected to do.
thanks for comments and thoughts on improving network disk storage performance.
1) intel x25-m ssd
2) wd5000aaks 500GB single drive
3) 4 wd5000aaks 500GB drives in RAID0 formation
4) apple time capsule (2TB, connected via gigabit ethernet)
i know that xbench is not the last word on performance, but since what mattered here was relative performance, i thought it was ok. i am currently using 10.6.1 on a mac pro.
results for the disk component of xbench?
1) intel ssd - 257.08 [amazing in random access]
2) single 500gb WD - 77.63
3) RAID0 - 229.84 [consistent between sequential and random]
4) time capsule - 16.82 [random performance was better than sequential, and bigger blocks yielded higher performance than smaller ones]
would you expect a single drive connected via gigabit ethernet (time capsule, in this case) to score only about 20% of an internal drive -- and about 7% of that of an SSD?
i guess the question comes down to how much gigabit ethernet's overhead is a constraint in terms of disk performance. i wonder if there are any setting i could change, or if it could simply be a constraint of the router on apple's time capsule, or if this is the best a single network drive could be expected to do.
thanks for comments and thoughts on improving network disk storage performance.