Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Locating a bad disk on Solaris w/Veritas

Re: Locating a bad disk on Solaris w/Veritas

From: Joel Garry <joel-garry_at_home.com>
Date: 27 Oct 2004 16:45:51 -0700
Message-ID: <91884734.0410271545.2a92a6e0@posting.google.com>


fitzjarrell_at_cox.net (David Fitzjarrell) wrote in message news:<9711ade0.0410270451.2c144c65_at_posting.google.com>...

>
> There is no compelling performance reason to separate tables and
> indexes in Oracle. This has been discussed many times in this
> newsgroup; one would hope the message is finally getting through.

Somehow I missed the part about how FTS with multiblock reads on a table loaded on a clean new disk isn't slower than one with index blocks interleaved in the TS (not to mention other table blocks). Could you point me directly to it please? Still haven't ever had time to prove it to myself one way or t'other. But it looks bad in Tablespace Map. Of course, compress=y looks good in Tablespace Map :-)

_Something_ makes my month-end batch type reports go faster when I imp/exp, I've just never quite figured out what. It's not due to deletion from indices, though, not doing that.

jg

--
@home.com is bogus.
Just what I need - keep my credit perfect, and lenders become
suspicious:  http://www.signonsandiego.com/uniontrib/20041024/news_1h24harney.html
Received on Wed Oct 27 2004 - 18:45:51 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US