Re: Ottawa - kernel hackers ...

[Alan Cox <alan redhat com>]

> 	Either way - the performance is directly related to seek time, and

Its related to two things - one is seek time the other is rotational 
delay (one of the reasons laptop disks suck more)

> almost nothing to do with read time. It's interesting that my hdparm
> measures read time, but not avg./worst case seek times - is there a good
> way for people to measure that easily - so we can correlate it with
> their times ?

bonnie and related tools are the normal disk tools. Hdparm's performance
tester is basically still there because people use it for "is my disk
working right" type tests.

bonnie is also cross platform so you can do meaningful linux/bsd/solaris/hpux
type comparisons too