Materialised View : Aggregate query involving partitionn operator - oracle

Question

Ran into a wall, trying to insert into Tableb with rows that meet a certain condition after insertion into Tablea
drop materialized view mv ;
drop materialized view log on tablea ;
create materialized view log on tablea
with rowid, sequence ( tino, price )
including new values;
create materialized view mv
refresh fast on commit
enable query rewrite
as
SELECT tino,sum(price)
FROM tablea PARTITION(PART_201609)
group by tino;
The above will return the ORA-12054 : cannot set ON COMMIT Refresh attribute.
Is this a limitation ? Aggregate query not having a partition operator in it?
Table is too larger, i would want my view to only have data specific to a certain period/month
When i remove the PARTITION(PART_201609) and ran as below i was able to successfully create the view :
create materialized view mv
refresh fast on commit
enable query rewrite
as
SELECT tino,sum(price)
FROM tablea
group by tino;
-- EDIT -- Include tablea's DDL
-- Create table
create table TABLEA
(
tino NUMBER not null,
price VARCHAR2(200),
dated DATE
)
partition by range (DATED)
(
partition PART_201608 values less than (TO_DATE(' 2016-09-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN')),
partition PART_201609 values less than (TO_DATE(' 2016-10-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN')),
partition PART_201610 values less than (TO_DATE(' 2016-11-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN')));

Answer 1

You can use the dbms_mview.explain_mview procedure to see why your proposed query can't be used for a fast-refresh-on-commit MV:
begin
dbms_mview.explain_mview(q'[SELECT tino,sum(price)
FROM tablea PARTITION(PART_201609)
group by tino]');
end;
/
select capability_name, possible, msgno, msgtxt from mv_capabilities_table;
CAPABILITY_NAME P MSGNO MSGTXT
------------------------------ - ---------- ------------------------------------------------------------------------------------------
PCT N
REFRESH_COMPLETE Y
REFRESH_FAST N
REWRITE Y
PCT_TABLE N 2067 no partition key or PMARKER or join dependent expression in select list
REFRESH_FAST_AFTER_INSERT N 2169 the materialized view contains partition extended table name
REFRESH_FAST_AFTER_ONETAB_DML N 2143 SUM(expr) without COUNT(expr)
REFRESH_FAST_AFTER_ONETAB_DML N 2146 see the reason why REFRESH_FAST_AFTER_INSERT is disabled
REFRESH_FAST_AFTER_ANY_DML N 2161 see the reason why REFRESH_FAST_AFTER_ONETAB_DML is disabled
REFRESH_FAST_PCT N 2157 PCT is not possible on any of the detail tables in the materialized view
REWRITE_FULL_TEXT_MATCH Y
REWRITE_PARTIAL_TEXT_MATCH Y
REWRITE_GENERAL N 2169 the materialized view contains partition extended table name
REWRITE_PCT N 2158 general rewrite is not possible or PCT is not possible on any of the detail tables
PCT_TABLE_REWRITE N 2185 no partition key or PMARKER in select list
As far as I'm aware there isn't any way around the the 2169 error:
02169, 00000, "the materialized view contains partition extended table name"
// *Cause: Fast refresh of materialized aggregate views and/or materialized
// join views are not supported if they were defined using partition
// extended table names.
// *Action: Create the fast refreshable materialized view without using
// partition extended table names or create the materialized view as
// a complete refresh materialized view.
Specifying the partition by name is somewhat unusual anyway; you can achieve the same thing by specifying the date range, and Oracle will limit the query to the relevant partition anyway. You get the same execution plan from:
explain plan for
select tino, sum(price)
from tablea partition(part_201609)
group by tino;
and
explain plan for
select tino, sum(price)
from tablea
where dated >= date '2016-09-01'
and dated < date '2016-10-01'
group by tino;
--------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop |
--------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 115 | 15 (7)| 00:00:01 | | |
| 1 | HASH GROUP BY | | 1 | 115 | 15 (7)| 00:00:01 | | |
| 2 | PARTITION RANGE SINGLE| | 1 | 115 | 14 (0)| 00:00:01 | 2 | 2 |
| 3 | TABLE ACCESS FULL | TABLEA | 1 | 115 | 14 (0)| 00:00:01 | 2 | 2 |
--------------------------------------------------------------------------------------------------
You'll see a higher row count than I get from my dummy table, but note the PSTART and PSTOP columns.
Using that for your MV still isn't quite enough though:
begin
dbms_mview.explain_mview(q'[select tino, sum(price)
from tablea
where dated >= date '2016-09-01'
and dated < date '2016-10-01'
group by tino]');
end;
/
select capability_name, possible, msgno, msgtxt from mv_capabilities_table;
CAPABILITY_NAME P MSGNO MSGTXT
------------------------------ - ---------- ------------------------------------------------------------------------------------------
PCT N
REFRESH_COMPLETE Y
REFRESH_FAST N
REWRITE Y
PCT_TABLE N 2067 no partition key or PMARKER or join dependent expression in select list
REFRESH_FAST_AFTER_INSERT N 2081 mv log does not have all necessary columns
REFRESH_FAST_AFTER_ONETAB_DML N 2143 SUM(expr) without COUNT(expr)
REFRESH_FAST_AFTER_ONETAB_DML N 2146 see the reason why REFRESH_FAST_AFTER_INSERT is disabled
REFRESH_FAST_AFTER_ONETAB_DML N 2142 COUNT(*) is not present in the select list
REFRESH_FAST_AFTER_ONETAB_DML N 2143 SUM(expr) without COUNT(expr)
REFRESH_FAST_AFTER_ANY_DML N 2161 see the reason why REFRESH_FAST_AFTER_ONETAB_DML is disabled
REFRESH_FAST_PCT N 2157 PCT is not possible on any of the detail tables in the materialized view
REWRITE_FULL_TEXT_MATCH Y
REWRITE_PARTIAL_TEXT_MATCH Y
REWRITE_GENERAL Y
REWRITE_PCT N 2158 general rewrite is not possible or PCT is not possible on any of the detail tables
PCT_TABLE_REWRITE N 2185 no partition key or PMARKER in select list
You need to resolve the 2067 error:
02067, 00000, "no partition key or PMARKER or join dependent expression in select list"
// *Cause: The capability in question is not supported when the materialized
// view unless the select list (and group by list if a GROUP BY
// clause is present) includes the partition key or
// PMARKER function reference to the table in question or an expression
// join dependent on the partitioning column of the table in question.
// *Action: Add the partition key or a PMARKER function reference or a join dependent
// expression to the select list (and the GROUP BY clause, if present).
... which is related to partition change tracking. You can add a partition marker to the select list and group-by, which again gets the same execution plan (PSTART/PSTOP), but now allows fast-refresh:
explain plan for
select dbms_mview.pmarker(rowid), tino, sum(price)
from tablea
where dated >= date '2016-09-01'
and dated < date '2016-10-01'
group by dbms_mview.pmarker(rowid), tino;
select * from table(dbms_xplan.display);
begin
dbms_mview.explain_mview(q'[select dbms_mview.pmarker(rowid), tino, sum(price)
from tablea
where dated >= date '2016-09-01'
and dated < date '2016-10-01'
group by dbms_mview.pmarker(rowid), tino]');
end;
/
CAPABILITY_NAME P MSGNO MSGTXT
------------------------------ - ---------- ------------------------------------------------------------------------------------------
PCT Y
REFRESH_COMPLETE Y
REFRESH_FAST Y
REWRITE Y
PCT_TABLE Y
REFRESH_FAST_AFTER_INSERT N 2081 mv log does not have all necessary columns
REFRESH_FAST_AFTER_ONETAB_DML N 2143 SUM(expr) without COUNT(expr)
REFRESH_FAST_AFTER_ONETAB_DML N 2146 see the reason why REFRESH_FAST_AFTER_INSERT is disabled
REFRESH_FAST_AFTER_ONETAB_DML N 2142 COUNT(*) is not present in the select list
REFRESH_FAST_AFTER_ONETAB_DML N 2143 SUM(expr) without COUNT(expr)
REFRESH_FAST_AFTER_ANY_DML N 2161 see the reason why REFRESH_FAST_AFTER_ONETAB_DML is disabled
REFRESH_FAST_PCT Y
REWRITE_FULL_TEXT_MATCH Y
REWRITE_PARTIAL_TEXT_MATCH Y
REWRITE_GENERAL Y
REWRITE_PCT Y
PCT_TABLE_REWRITE Y
And you can indeed use that query to create your MV:
create materialized view mv
refresh fast on commit
enable query rewrite
as
select dbms_mview.pmarker(rowid), tino, sum(price)
from tablea
where dated >= date '2016-09-01'
and dated < date '2016-10-01'
group by dbms_mview.pmarker(rowid), tino;
Materialized view MV created.
If you want to enable all capabilities in the MV you can add the dated column to your MV log:
create materialized view log on tablea
with rowid, sequence ( dated, tino, price )
including new values;
and include the missing aggregates in your MV query:
select dbms_mview.pmarker(rowid), tino, sum(price), count(price), count(*)
from tablea a
where dated >= date '2016-09-01'
and dated < date '2016-10-01'
group by dbms_mview.pmarker(rowid), tino
Not relevant, but also note that you can, if it's beneficial, partition the MV view log too:
create materialized view log on tablea
partition by range (dated)
(
partition PART_201608 values less than (TO_DATE(' 2016-09-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN')),
partition PART_201609 values less than (TO_DATE(' 2016-10-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN')),
partition PART_201610 values less than (TO_DATE(' 2016-11-01 00:00:00', 'SYYYY-MM-DD HH24:MI:SS', 'NLS_CALENDAR=GREGORIAN'))
)
with rowid, sequence ( dated, tino, price )
including new values;