Order of keys inside index blocks
In this post we are going to illustrate how index keys are ordered inside a leaf block of an Oracle B+tree index.
It is well known that index blocks share most of their structure with “regular” (heap) table blocks; in particular, they share most of the way entries are stored inside the block. In tables, a row can be placed anywhere in the bottom part of the block (which is essentially managed like an “heap”, in which the exact memory address that the row is stored at is not important); its row address is recorded into one of the slots of a vector named “row directory”, located near the beginning of the block.…
