On Thu, 4 Mar 2004, Paul Thomas wrote:
> On 03/03/2004 18:23 scott.marlowe wrote:
> > [snip]
> > There are three factors that affect how fast you can get to the next
> > sector:
> >
> > seek time
> > settle time
> > rotational latency
> >
> > Most drives only list the first, and don't bother to mention the other
> > two.
>
> Ah yes, one of my (very) few still functioning brain cells was nagging
> about another bit of time in the equation :)
>
> > On many modern drives, the seek times are around 5 to 10 milliseconds.
> > [snip]
>
> Going back to the OPs posting about random_page_cost, imagine I have 2
> servers identical in every way except the disk drive. Server A has a 10K
> rpm drive and server B has a 15K rpm drive. Seek/settle times aren't
> spectacularly different between the 2 drives. I'm wondering if drive B
> might actually merit a _higher_ random_page_cost than drive A as, once it
> gets settled on a disk track, it can suck the data off a lot faster.
> opinions/experiences anyone?
It might well be that you have higher settle times that offset the small
gain in rotational latency. I haven't looked into it, so I don't know one
way or the other, but it seems a reasonable assumption.
However, a common misconception is that the higher angular velocity of
the 15krpm drives would allow you to read data faster. In fact, the limit
of how fast you can read is set by the head. There's a maximum frequency
that it can read, and the areal density / rpm have to be such that you
don't exceed that frequency. OFten, the speed at which you read off the
platters is exactly the same between a 10k and 15k of the same family.
The required lower areal density is the reason 15krpm drives show up in
the lower capacities first.