Parallel Booting
Jason L Tibbitts III
tibbs at math.uh.edu
Sat Jul 7 01:26:07 UTC 2007
>>>>> "AC" == Alan Cox <alan at redhat.com> writes:
AC> Disk performance is usually quantified in ops/second. If you read
AC> linear data you'll get 57MB/sec. Go to true random access and
AC> performance is way way lower and hasn't improved much in the past
AC> ten years.
Well, these aren't disks (essentially zero seek time), so I'd expect
random access read performance to be extremely fast. Random writes,
on the other hand, should be terrible. I didn't have a lot of time to
benchmark this before heading home so I just ran bonnie++ but I left
off some options and the second line of results was entirely filled
with plusses.
But anyway, I was merely responding to the comment about SSDs
rendering this discussion moot. Obviously disk access speed is a
factor in startup time, but I don't think it's anything close to the
dominating factor.
- J<
More information about the devel
mailing list