Oracle 性能优化实战

yyp2009

原帖由 nj21 于 2011-4-8 14:50 发表
我修改了一下，鬼兄看看对不对？

假设db_file_multiblock_read_count是10，而表中有100条记录，分布在10个block上，highwater mark也就在这里，现在要拿到这100条记录，用全表扫描，只需要读一次；而用index scan，则至少200次block accesses。

出处在这里，不知道我理解得对不对？
What this means is that for each row that will be retrieved via an index scan, at least two block
accesses will be required: at least one index block and one data block. If your final result set contains
100 rows and those 100 rows are retrieved using an index scan, there would be at least 200 block
accesses required.

请问一下楼主这句话话是你哪里看来的呀？
What this means is that for each row that will be retrieved via an index scan, at least two block
accesses will be required: at least one index block and one data block. If your final result set contains
100 rows and those 100 rows are retrieved using an index scan, there would be at least 200 block

这句话？

[ 本帖最后由 yyp2009 于 2011-4-9 22:37 编辑 ]

xin1shou3

谢谢！

nj21

这段话来自于
Pro Oracle SQL

zhanglikui

受教了