Supporting Hardware - Expert Consolidation in Oracle Database 12c - page 107

The above output shows the AWR report for an execution of runit.sh 16 and a 16MB buffer cache. This
observation is confirmed by collectl, too, which has been running in parallel:
#<----CPU[HYPER]-----><---------------Disks---------------->
#cpu sys inter ctxsw KBRead Reads Size KBWrit Writes Size
2 1 42236 47425 182168 22771 8 0 0 0
2 1 43790 47359 182272 22776 8 16 1 16
2 1 42955 46644 179040 22380 8 24 3 8
Which processing lead to those numbers? So far it has not been described what the benchmark does in more
detail. The logic is defined in the slob.sql file. The new file combines the application logic, which has been defined
in two different files, readers.sql and writers.sql. The new file follows the same model.
For a read-only-workload, in which UPDATE_PCT is set to 0 in slob.conf, you will see select statements
exclusively similar to this one taken from the AWR report:
SQL> select * from table(dbms_xplan.display_cursor('309mwpa4tk161'))
PLAN_TABLE_OUTPUT
-----------------------------------------------------------------------------------------------
SQL_ID 309mwpa4tk161, child number 0
-------------------------------------
SELECT COUNT(C2) FROM CF1 WHERE CUSTID > ( :B1 - :B2 ) AND (CUSTID <
:B1 )
Plan hash value: 1497866750
------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | | | 259 (100)| |
| 1 | SORT AGGREGATE | | 1 | 133 | | |
|* 2 | FILTER | | | | | |
| 3 | TABLE ACCESS BY INDEX ROWID BATCHED | CF1 | 256 | 34048 | 259 (0)| 00:00:01 |
|* 4 | INDEX RANGE SCAN | I_CF1 | 256 | | 2 (0)| 00:00:01 |
------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - filter(:B1>:B1-:B2)
4 - access("CUSTID">:B1-:B2 AND "CUSTID"<:B1)
23 rows selected.
As you can see, the reader part of the script will perform an index range scan. It is advisable to tail the database's
alert log when creating the data set, as well as when you are running the benchmark, to ensure that you are not
running out of space or other system resources, such as processes.