Re: version_count

On Apr 4, 9:38 am, Chuck <skilover_nos..._at_bluebottle.com> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Can someone please give me an explanation, or point me to a URL which
> has one, of what v$sqlarea.version_count is all about? I understand it
> has to do with the # of child cursors for a given SQL in the library
> cache, but why would you need more than one cursor? Is that cursor not
> shareable? What would make a cursor not shareable? Does it have to do
> with different execution plans for the same SQL but with different bind
> variables? Or is it simply the # of sessions running the same SQL at he
> same time? TIA
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.5 (GNU/Linux)
> Comment: Using GnuPG with PCLinuxOS -http://enigmail.mozdev.org
>
> iEYEARECAAYFAkf2WZAACgkQzIf+rZpn0oTspACdFxt0gFkvXgtJ+kMh3oANr0s7
> nwkAoJ8SS36nI0QHWoprM7ZMlSG3dhk+
> =GE+T
> -----END PGP SIGNATURE-----
From the performance manual: "A SQL statement can map to multiple cursors, because the objects referred to in the cursor can differ from user to user."

There's lots of explanations on the web and in books. Here's a good start (found by searching for v$sqlarea.version_count on asktom.oraclecom) http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:9497064796920 http://www.freelists.org/archives/oracle-l/03-2004/msg03180.html

Dan's primer is pretty good summary of 11: http://www.psoug.org/reference/cursor_sharing.html

A view exists to tell you why:
http://download.oracle.com/docs/cd/B19306_01/server.102/b14237/dynviews_2124.htm

jg

--
@home.com is bogus.
That darn internet.  http://www.ic3.gov/media/2008/080403.htm