index full scan 和 index FAST full scan 區別

index full scan starts at the root block, walks down the branches on the left
hand side one by one till it gets to the first leaf block on the "left" and then
reads the index one block at a time -- leaf by leaf.  it reads the data
"sorted".

index FAST full scan reads the extent map for the index and using multi-block IO
reads the index in big chunks.  It throws out all root/branch blocks and
processes the data it finds on the leaf blocks.  It reads the data "in some
random order".


So a FAST full scan would generally read the index faster due to the multiblock
IO.

好帖，學習一下

帖子原文出處： http://www.itpub.net/thread-346360-2-1.html