Materialised view construct

Question

Hi all,
I have some queries that do a lot of calculations based on year passed from front-end.
I am unable to figure out how could i use a materialised view so that calaculations should be done on current data rather than the entire data.
As calculations are done based on date passed from front-end,How to convert such queries to a materialised view?
Here is a sample query :
select mjcd,desc1,
bm.amt,
nvl(sum(case when to_char(v_date,'RRRR')||
    lpad(to_char(v_date,'MM'),2,'0')='200710'
     then gross_amt end),0)monthly_amt,
nvl(sum(gross_amt),0)progressive
   from v_billent be,mjhd mj
,v_bm bm
   where v_date>='01-MAR-07' and v_date<='30-OCT-07'
   and mjcd>='2011'
and bm.fin_year='20072008'
  and mj.mjcd=be.src_mjcd(+)
  and bm.h_code(+)=mj.mjcd
   group by mjcd,desc1
,bm.amtorder by 1
Kindly guide how can I use advantage of a materialised view?
Thanks

Answer 1

Materialized view are greatly used to increase the performance of pre existing aggregate queries.
See a small example below.
SQL> create table t
  2  as
  3  select * from all_objects
  4  /
Table created.
SQL> exec dbms_stats.gather_table_stats('SYSADM','T')
PL/SQL procedure successfully completed.Now i want to execute a query like this.
SQL> explain plan for
  2  select object_type, count(1)
  3    from t
  4  group by object_type
  5  /
Explained.
SQL> set linesize 250
SQL>
SQL> select * from table(dbms_xplan.display)
  2  /
PLAN_TABLE_OUTPUT
Plan hash value: 47235625
| Id  | Operation          | Name | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT   |      |    21 |   168 |   374   (7)| 00:00:05 |
|   1 |  HASH GROUP BY     |      |    21 |   168 |   374   (7)| 00:00:05 |
|   2 |   TABLE ACCESS FULL| T    |   117K|   914K|   358   (2)| 00:00:05 |
9 rows selected.Lets say over a period of time the data in table T has increased so much and
i am facing a big performance problem. so to over come that i create a Materialized
view.
SQL> create materialized view t_mv
  2  enable query rewrite
  3  as
  4  select object_type, count(1)
  5    from t
  6  group by object_type
  7  /
Materialized view created.Now when is issue the same query see what happens
SQL> delete from plan_table
  2  /
4 rows deleted.
SQL> explain plan for
  2  select object_type, count(1)
  3    from t
  4  group by object_type
  5  /
Explained.
SQL> select * from table(dbms_xplan.display)
  2  /
PLAN_TABLE_OUTPUT
Plan hash value: 139569370
| Id  | Operation                    | Name | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT             |      |    31 |   744 |     3   (0)| 00:00:01 |
|   1 |  MAT_VIEW REWRITE ACCESS FULL| T_MV |    31 |   744 |     3   (0)| 00:00:01 |
Note
PLAN_TABLE_OUTPUT
   - dynamic sampling used for this statement
12 rows selected.
SQL>instead of picking from my base table T it goes for the materialized view T_MV.
And check out the cost it got reduced drastically.
There is lot more than what i have said about materialized view. You can read the doc
for more info.