Re: FULL TABLE SCANS...

There is nothing wrong with controlled full table scans - they can be much faster than index scans. Full table scans read a number of blocks a time (init .ora parameter multiblock read count) if say 8 blocks are read ech containg on average 10 rows then 1 read gives 80 rows. an 80million row table will require 1 million reads. Reading the entire table using an index will require reading the index blocks and then for every row, probably going to a data block and probably doing an io i.e 80 million i/os in the example. The bottom line is that if you want more than 20% of a table the probably a full table scan is more efficient, there are a number of caveats, it depends upon rows per block, degrees of parallism, is the data you need stored entirely in the index. The time you dont want full table scans are when just a few rows are needed.

    Nick

Michael Rosadino <mrosadino_at_connico.net> wrote in article <34668421.3162_at_connico.net>...
> Here's a quick question which I know probably has a very long answer.
Received on Mon Nov 17 1997 - 00:00:00 CET