How indexes are organized : LUSENET : DBAzine : One Thread

I want to find out more about how B*tree indexes are organized within Oracle and how they work. For example, when I create an index on multiple columns, what data is stored in the index blocks ? When does a block split occurs ? What exactly are the contents of a root block and node blocks ? When does a block become a root block or a node block ? When the index data cannot be contained in one block, what happens ? Can I treat an index as a two column table ? Can I look at the index data ? Since indexes store the rowid, where do I take the rowid size into account while index sizing ? After a block split occurs, does my index sizing calculation hold good for the node blocks or the root block ?

Thanks much... Arun

-- Arun Gupta (, April 16, 2002

Moderation questions? read the FAQ