Querying data from the buffer_cache

Question

Hi,
I'm running an experiment on my own home database to understand how data is retrieved from the buffer_cache. Using Oracle 10g 10.2.0.3.0. I've populated 2 tables with 10million records each. The table structure is as follows.
create table bigtable(col1 varchar2(50), col2 varchar2(50))
create table bigtablechild(col1 varchar2(50), col2 varchar2(50))
bigtablechild.col1 is a foreign key to bigtable.col1 and there are no indexes on the tables.
When I run this query, it takes about 30 s to return data each time it runs. I can also see hard drive activity running on my handy hard drive monitor utility each time. I thought that once the data is loaded into the buffer_cache, the query wouldn't scan the hard disk for data. Could someone please help me understand what is happening?
select b.col2 from bigtable a, bigtablechild b where a.col1 = b.col1 and a.col1 = 'ABC8876'
| Id  | Operation          | Name          | Starts | E-Rows | A-Rows |   A-Time   | Buffers | Reads  |
|*  1 |  HASH JOIN         |               |      1 |      5 |      5 |00:00:23.00 |   93671 |  90663 |
|*  2 |   TABLE ACCESS FULL| BIGTABLE      |      1 |      1 |      1 |00:00:14.26 |   57799 |  54931 |
|*  3 |   TABLE ACCESS FULL| BIGTABLECHILD |      1 |      5 |      5 |00:00:08.74 |   35872 |  35732 |
-------------------------------------------------------------------------------------------------------Edited by: arizona9952 on Jan 19, 2013 9:38 AM

Answer 1

It's not the o/s RAM that would be used for the caching of the data but the amount of RAM that youhave designated as buffer cache within your SGA , that would be used for it. You are probably running the tests on large tables and you don't have enough sized buffer cache. It's correct that with the buffer cache, next time, the access is done from the buffer cache provided it holds the data. Using some small tables you can see this behavior like below,
SQL*Plus: Release 11.2.0.3.0 Production on Sat Jan 19 22:48:14 2013
Copyright (c) 1982, 2011, Oracle.  All rights reserved.
startConnected to an idle instance.
SQL> startup
SP2-0734: unknown command beginning "startstart..." - rest of line ignored.
SQL> startup
ORACLE instance started.
Total System Global Area  263639040 bytes
Fixed Size                  1344312 bytes
Variable Size             205524168 bytes
Database Buffers           54525952 bytes
Redo Buffers                2244608 bytes
Database mounted.
Database opened.
SQL> set autot trace stat
select            
SP2-0042: unknown command "s" - rest of line ignored.
SQL> select * from scott.emp;
14 rows selected.
Statistics
         57  recursive calls
          0  db block gets
         87  consistent gets
         11  physical reads
          0  redo size
       1454  bytes sent via SQL*Net to client
        419  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          6  sorts (memory)
          0  sorts (disk)
         14  rows processed
SQL> /     
14 rows selected.
Statistics
          0  recursive calls
          0  db block gets
          7  consistent gets
          0  physical reads
          0  redo size
       1454  bytes sent via SQL*Net to client
        419  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
         14  rows processedYou can see that the next time when the query is run, there are no physical IO's but only logical ones.
HTH
Aman....