Re: dc_used_extents ,dc_free_extents and dc_histogram_defs

[Denny Koovakattu]

And how do you join one ? Denny Pete Sharman wrote: Well, you could have joined the beta program if you were THAT interested, Mladen! Ducks and runs. :) Pete "Controlling developers is like herding cats." Kevin Loney, Oracle DBA Handbook "Oh no, it's not. It's much harder than that!" Bruce Pihlamae, long-term Oracle DBA -----Original Message----- Mladen Gogala Sent: Tuesday, December 09, 2003 6:59 AM To: Multiple recipients of list ORACLE-L Larry Ellison has publicly stated that his goal is to produce a database with less then 100 tunable parameters. Allegedly, he came rather close with 10g. As far as 10g is concerned, I'm rather disappointed with the marketing hype being created with oracle not making an early version available. I don't plan on migrating to 10g until I learn it well and if some oracle sales guy tries to exert pressure on me to migrate, he will get a very stable sign used by English archers after the battle at Agincourt to signify that they still have all the fingers needed to operate a longbow. I've had my fill of white papers and articles and now I want to see the software. On 12/08/2003 02:24:33 PM, [EMAIL PROTECTED] wrote: .. And there used to be all these dc_ parameters that one could set, giving the dba control over the dictionary cache, which was not a part of the shared pool. And then came Oracle V7, with the shared_pool_size, wresting that control. Regards Mladen Gogala <[EMAIL PROTECTED] To: Multiple recipients of list ORACLE-L <[EMAIL PROTECTED]> ading.com> cc: Sent by: Subject: Re: dc_used_extents ,dc_free_extents and dc_histogram_defs [EMAIL PROTECTED] ity.com 12/08/2003 01:29 PM Please respond to ORACLE-L Well, once upon a time, in a land far, far away, there was Oracle V6 with something called "TPO", which was essentially row locking + PL/SQL V1 (no stored procedures). My guess is that Cary Millsap, Anjo Kolk, Steve Feuerstein and Howard Rogers know a thing or two about the Jurassic period in the database development. BTW, that was also when buffer hit ratio was invented. The entries that you see are remnants from oracle v6, together with the table called "V$ROWCACHE" and are both religiously maintained for the compatibility reasons, because Oracle Corp. doesn't want to disappoint all those who are still running V6. Even compatibility with V5 is still maintained. In Oracle 5.1.22, dictionary views weren't called "user_tables" and "user_objects", they were called "tab" (user_tables) and cat (from "CATALOG", replaced with "USER_OBJECTS"). The term "CATALOG" was directory command on Apple IIe (6502, later Z80) with 100k floppies, computer immensly popular at the time, and I believe that is why the first implementation of "user_objects" was called "catalog". Now, let's fast forward to the present time and Oracle 9.2.0.4. Do "Select * from tab" and "select * from cat". You'll be surprised. For all those still running V5.1.22 with forms 2.0 and 2.3, the world is not over yet. On 12/08/2003 12:39:30 PM, Guang Mei wrote: Hi: I am reading some statspack reports from our 8173 DB (on Sun Solaris) and found some of "Dictionary Cache Stats" are pretty high (much higher than 2%). I notice that "Pct Get Miss" for dc_used_extents ,dc_free_extents and dc_histogram_defs are high (the second column data below). Is this something I need to pay attention in terms of doing performance optimization? If yes, what are the things (regarding "Dictionary Cache") that I should look in order to improve the performance? TIA. Guang ps, here are some "dc_" stats from my reports and a copy of actual report (partial):