Tag: oracle database performance enhancement

Enhancing Oracle Database System IO Performance: filesystemio_options Parameter

arcsdegeoLeave a comment

There is an Oracle database parameter filesystemio_options that I found out most DBA’s are not aware of.  First, this parameter can be used if your environment is NOT based on ASM setup.

In default behaviour…In buffered I/O, the Operating System maintains its own cache of disk data. Rather than directly reading to or writing from a process buffer, data is read from the disk into the cache and copied to the process buffer or copied from the process into the cache and written from there to the disk. Moreover, read requests are processed through cached data without having to read it again from the disk and can prefetch data from the disk into the cache before processes request it, speeding reads for data.

The catch here is that Oracle database system is already has in-place buffer cache for (read,write) to disk operations, so there will be “double” operations going on from DB & OS and will cause CPU performance overhead. However, direct IO will bypass the double buffering overhead.

Add the following parameter in init.ora and re-start the database instance:

filesystemio_options=setall

To observe the performance boost, generate AWR report before setting the parameter ( period of 1 week) and compare it after setting the parameter for 1 week report generation…..and the findings:

BEFORE PARAMETER SET:

AFTER PARAMETER SET:

DB CPU, db file sequential read, direct path read wait events are drastically improved as shown in BEFORE/AFTER figures.

Oracle Database In-Memory Base Level Feature in Oracle 19c

arcsdegeoLeave a comment

In-Memory database technology is already introduced in Oracle database system, however in Oracle 19c things changed ! The feature is now free with limited size of  16 GB. You can now explore “Base-Level” feature starting with Oracle 19.8 version where JULY 2020 RU (Release Update) is applied.

The new Base Level feature supports all Database In-Memory features except for the following:

  • Automatic In-Memory  (AIM)
  • Compression levels other than MEMCOMPRESS FOR QUERY LOW
  • Excluded columns (all columns of a table are populated)
  • The CELLMEMORY feature on Exadata

To enable In-Memory “Base-level” feature:

go to cd $ORACLE_HOME/dbs and edit your init.ora file by adding the following two parameters and “restart” your database

INMEMORY_FORCE=BASE_LEVEL

INMEMORY_SIZE=7G

SQL> create table dba.in_mem as select * from dba_objects;

Table created.

SQL> explain plan SET STATEMENT_ID=’EMO1′ for select count(*) from dba.in_mem;

SQL> SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY(‘PLAN_TABLE’,’EMO1′,’ALL’));

PLAN_TABLE_OUTPUT

——————————————————————————–

Plan hash value: 3285478656

———————————————————————

| Id | Operation | Name | Rows | Cost (%CPU)| Time |

———————————————————————

| 0 | SELECT STATEMENT | | 1 | 115 (0)| 00:00:01 |

| 1 | SORT AGGREGATE | | 1 | | |

| 2 | TABLE ACCESS FULL| IN_MEM | 25010 | 115 (0)| 00:00:01 |

———————————————————————

Query Block Name / Object Alias (identified by operation id):

PLAN_TABLE_OUTPUT

——————————————————————————–

————————————————————-

1 – SEL$1

2 – SEL$1 / IN_MEM@SEL$1

Column Projection Information (identified by operation id):

———————————————————–

1 – (#keys=0) COUNT(*)[22]

2 – (rowset=1019)

21 rows selected.

*** — Now i will load the table to Memory:

SQL> ALTER TABLE dba.in_mem INMEMORY PRIORITY HIGH;

Table altered.

— to check objects uploaded in memory you can query dictionary view V$IM_SEGMENTS:

SQL> select OWNER,SEGMENT_NAME,SEGMENT_TYPE from v$IM_SEGMENTS ;

OWNER SEGMENT_NAME SEGMENT_TYPE

——————————————————————————————————————————– ——————————————————————————————————————————– ——————

dba IN_MEM TABLE

SQL> explain plan SET STATEMENT_ID=’EMO2′ for select count(*) from dba.in_mem;

SQL> SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY(‘PLAN_TABLE’,’EMO2′,’ALL’));

PLAN_TABLE_OUTPUT

————————————————————————————————————————————————————————————————————————————————————————————————————

Plan hash value: 3285478656

——————————————————————————

| Id | Operation | Name | Rows | Cost (%CPU)| Time |

——————————————————————————

| 0 | SELECT STATEMENT | | 1 | 5 (0)| 00:00:01 |

| 1 | SORT AGGREGATE | | 1 | | |

| 2 | TABLE ACCESS INMEMORY FULL| IN_MEM | 25010 | 5 (0)| 00:00:01 |

——————————————————————————

Query Block Name / Object Alias (identified by operation id):

————————————————————-

1 – SEL$1

2 – SEL$1 / IN_MEM@SEL$1

Column Projection Information (identified by operation id):

———————————————————–

1 – (#keys=0) COUNT(*)[22]

2 – (rowset=1019)

21 rows selected.

 

as shown above the explain plan shows full scan against database table stored in-memory.

** resources/refrences:

https://blogs.oracle.com/in-memory/base-level-198-ru

https://docs.oracle.com/en/database/oracle/oracle-database/20/ftnew/database-memory-base-level.html

https://docs.oracle.com/en/database/oracle/oracle-database/19/refrn/INMEMORY_FORCE.html#GUID-1CAEDEBC-AE38-428D-B07E-6718A7225548

https://www.oracle.com/a/tech/docs/twp-oracle-database-in-memory-19c.pdf