Re: Buffer cache statistics (ratios) and CBO SQL optimization?

Given the lateness of this reply and the context of its thread, I think it's more than high time Geomancer identified him/herself to the group, if not in email at least in real name. There are remarkably few here that use a real e-mail, but almost all identify themselves with their real names, enough that we can readily id them.

It's highly suspicious how his/her replies defending some books generate interventions from the authors with monotonous regularity. Particularly when such authors have seldom contributed here before. If ever. No, it's not a "don't have time" thing either!

Let me be clear: nothing wrong IMO with such authors coming here to talk to everyone, if properly identified and in context. Most welcome, and frequently. Plenty of others do the same (JL, SA, CMD, HJR, etc) regularly, and correctly identifying themselves. Super.

PLENTY wrong with "moles", no matter how sweet their talk might be. And starting to get tiring and irritating. Let's hope adult reasoning and the most basic HONESTY prevails: there are precious few people in this ng that haven't been there seen it, when it comes to childish scams.

-- 
Cheers
Nuno Souto
wizofoz2k_at_yahoo.com.au.nospam
"Dave Moore" <dave_at_oracleutilities.com> wrote in message news:dcfe458b.0401070602.364c5b26_at_posting.google.com...


> As the author of the Oracle utilities book, I can assure you it's not

> a cut and paste from OTN. The only reason I wrote the book was to

> share insight that was orginal and that I couldn't find anywhere else.

>  For instance, the book includes how to actually use orakill in

> combination with something like QSlice ... or how to create your own

> facility and acutally make oerr useful.  Trace Analyzer was explained

> as well as scripts provided that aid in the tracing of user sessions

> (all Oracle tracing tools were detailed).  For some of the common

> utilities (imp, exp, sqlldr) I discussed performance and gave the

> benchmarks when using many different options on a table with 1M rows.

> Should you use exp/imp or the SQL*Plus copy command? It depends, but

> many times the copy command will do it even faster and with less

> hassles.

>

> I couldn't dive straight into the "value add" part of each utility

> until I explained first what it did.  Some people are not even aware

> of utilities like maxmem or oradebug.  For those, I took a different

> approach - what does it do, how do I execute it, and what problem can

> it solve for me. Hope this helps.

>

> Dave Moore