Re: should initial = next?

On Wed, 07 Feb 2001 09:08:24 +1030, Steve Salvemini <steve.salvemini_at_adelaide.edu.au> wrote:

>I saw a another comment by Dave Ensor:
>
>"Your goal is to minimize the number of extents on disk. Access to
>contiguous areas of disk is much faster than access to noncontiguous
>areas. In one test that we did on a 4,000-row table, we found that when
>the entire table fit on one extent, it took 0.76 second to scan it; when
>the table was spread over 10 extents, it took 3.62 seconds.
>

One thing to keep in mind is the context of this. In most systems nowadays, you have things like EMC or equivalent controllers. The argument for coniguous areas on disk goes totally out the window with this type of hardware. It just doesn't mean anything given that they behave as a black box as far as disk allocations go. What Dave says is still valid to an extent(!) for systems with direct disk connections, via conventional controllers.

Cheers
Nuno Souto
nsouto_at_bigpond.net.au.nospam
http://www.users.bigpond.net.au/the_Den/index.html Received on Wed Feb 07 2001 - 04:12:42 CST