RE: 11gR2 - what patch level?

Not sure if this is the kind of thing you're looking for (since it's not strictly upgrade related), but a new underscore init param was introduced in 11.2 called _optimizer_join_factorization.

It allows the optimizer to transform statements like (t1 join t2) union (t1 join t3)
Into
t1 join (t2 union t3)
(see :
http://timurakhmadeev.wordpress.com/2009/12/01/join-factorization/)

It defaults to true and of course already has a couple of associated bugs. The one that bit us pretty badly was :

Bug 8692170 SubOptimal Execution Plan with Join Factorization enabled.

"suboptimal" in this case being an understatement. No prizes for guessing what the Oracle Support-proposed solution was.

mvg / regards,
Jo Holvoet    

-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Nuno Souto Sent: zaterdag 3 juli 2010 6:46
To: Mark W. Farnham
Cc: oracle-l_at_freelists.org
Subject: Re: 11gR2 - what patch level?

Thanks heaps, Mark. I'm aware of the differences you mentioned, we have some
dbs in 11r1 and those were ironed out a while ago. I'm more interested in the
so-called "gotchas".
Like the resource manager being left on by default in r2 and other such. As
well as any known bugs that may have affected other's upgrades. Been reading the
patch bug fixes and those are always very helpful. But nothing replaces talking
to the "coal face" folks.
We don't use RAC or ASM so we are more or less shielded from the problems
associated with those. But we do use roles and login triggers extensively to
control access and environment for multiple applications in same db. And PL/SQL
to a great extent. As well as standby, via SAN replication. Ah yes: we're on
IBM's AIX, power6.

-- 
Cheers
Nuno Souto
in cold Sydney, Australia
dbvision_at_iinet.net.au


Mark W. Farnham wrote,on my timestamp of 3/07/2010 12:21 PM:


> Non-exhaustive list starter:

> 

> 1) case sensitivity of passwords (which you can switch off, but which

is on


> by default, may wreak havoc if you have passwords laying around

encrypted


> somewhere lowercase for programs to pick up or in root guarded scripts

but


> the real existing passwords are really uppercase unless someone

defined them


> with surrounding quotes. Now it matters what you toss at the login

software.


> 

> 2) changes (many permissions removed) for the connect privilege. They

have


> provided stuff to handle it and recommendations about it, but you need

to


> notice that you probably need to do something. I think the new

defaults are


> improvements to security, but there is work for you to do to make your

> existing least privileged existing connectors able to do about

anything.


> 

> y'all chime in.


--
http://www.freelists.org/webpage/oracle-l


--
http://www.freelists.org/webpage/oracle-l