Jonathan Lewis wrote a series of posts on topic of Index ITLs (interested transaction list), how and when they become huge and make index bigger in size – see this post with links to several other articles. As it turned out in a recent thread on SQL.ru forum, one of possible solutions to the issue – calling ALTER index COALESCE – appears to be very limited in it’s appliance. Why? Because COALESCE doesn’t clean out ITLs of all index leaf blocks – most important, it doesn’t clean out ITL of the right-most index leaf block.
-
Categories
-
Tags
10.2.0.3 10.2.0.4 10.2.0.5 10g 10gR2 11.2.0.1 11.2.0.2 11.2.0.3 11g 11gR2 ANSI SQL ASH AWR benchmark bind peeking bug bug-or-feature CBQT coalesce coalesce subquery collections conferences connect by control files docs dynamic views EBR Exadata fail FBI Firefox full join funny stuff Hotsos indexes iPX ITL javelin JNI join elimination join factorization memory merge MOS muxers nested loops OIC OICA OOM optimizer_features_enable opt_param ORA-04031 Oracle JVM outer join PDML poll presentations PX RAC scripts shared cursors SPECjAppServer2004 sql_trace statistics Statspack subqueries subquery factoring subquery unnesting Sun Tom Kyte tpt trace files undocumented wait events wish list -
Top Posts
-
Blogroll
-
Join 529 other subscribers