Comments on: Index Explosion 3

By: Glossary « Oracle Scratchpad

Glossary « Oracle Scratchpad — Sat, 20 Feb 2010 18:50:40 +0000

[...] maxtrans. (From 10g, Oracle has stared to ignore the setting of maxtrans, and this can cause some space wastage problems in indexes). The initial size of the ITL is set by initrans- with a minimum setting that is dependent on the [...]

By: Jonathan Lewis

Jonathan Lewis — Sat, 29 Aug 2009 15:08:25 +0000

Martin,

Sorry, I was a little sloppy with language. The “flush shared_pool” command probably deserves to be changed to “flush sga_heap”, as it attempts to clear various structures in the sga heap, including the library cache and dictionary cache.

Since the flush command doesn’t always do the job properly on the library cache, I take the pessimistic viewpoint that it might not work perfectly on the dictionary cache as well.

It’s quite possible that none of the flushes is necessary, and a little more likely that just the first is sufficient.

By: Log Buffer #159: a Carnival of the Vanities for DBAs | Pythian Group Blog

Log Buffer #159: a Carnival of the Vanities for DBAs | Pythian Group Blog — Fri, 28 Aug 2009 17:05:58 +0000

[...] the Oracle Scratchpad, Jonathan Lewis has the third part of his Index Explosion series, this time answering the quesiton, “How can you set maxtrans on an index in 10g (and [...]

By: Martin Berger

Martin Berger — Fri, 28 Aug 2009 06:33:11 +0000

Jonathan,
I didn’t get the reason for the 2 flushes of shared pool: You argued, in some versions it will not flush the library cache properly at the 1st time. But, you only want to avoid the dictionary cache to write back to ind$. How is this affected by library cache? I fear, your steps are too big and I missed the trace. Maybe you can add an additional guidepost for me?
thank you,
Martin