Each time an ETL job finished I have to collect the actual used space for each table (including its index spaces), as well as the row count.
The bytes column from user_segments and user_extents only indicates the space allocated. So I used the following to get each table size
Dbms_Space.object_space_usage (
object_owner => v_owner,
object_name => c.name,
object_type => c.typ,
sample_control => NULL,
space_used => v_space_used_out,
space_allocated => v_space_allocated_out,
chain_pcent => v_chain_pcent_out);
v_space_used := v_space_used+v_space_used_out;
v_space_allocated := v_space_allocated+v_space_allocated_out;
looping through each indexes and the table itself.
My question is before I run the above code is it necessary to run?
EXEC DBMS_STATS.GATHER_TABLE_STATS(v_owner,v_table_name);
Why and/or why no?
Thanks,
It is not necessary to run EXEC DBMS_STATS.GATHER_TABLE_STATS(v_owner,v_table_name) ahead of running DBMS_SPACE.OBJECT_SPACE_USAGE
I ran Dbms_Space.object_space_usage, got the results. And then inserted a lot of data into a table, and ran Dbms_Space.object_space_usage a second time.
The results were updated to the new, larger values, without running GATHER_TABLE_STATS.
We know from Oracle DBMS_SPACE documentation that this package requires the analyze privilege. However, merely running OBJECT_SPACE_USAGE on a table, does not update the last_analyzed column of all_tables.
So from observation, DBMS_SPACE.OBJECT_SPACE_USAGE appears to do its own analysis of the table, getting correct space usage. It does not rely on updated stats being collected by DBMS_STATS nor does it update the stats.
Just to further prove the point:
SQL> EXEC DBMS_STATS.delete_table_stats('SOMEOWNER','SOMETABLE');
PL/SQL procedure successfully completed.
SELECT LAST_ANALYZED
FROM ALL_TABLES
WHERE OWNER = 'SOMEOWNER' AND TABLE_NAME = 'SOMETABLE';
LAST_ANALYZED
------------------
SQL>
DECLARE
su NUMBER;
sa NUMBER;
cp NUMBER;
BEGIN
dbms_space.object_space_usage('SOMEOWNER', 'SOMETABLE', 'TABLE',
NULL, su, sa, cp);
dbms_output.put_line('Space Used: ' || TO_CHAR(su));
dbms_output.put_line('Space Allocated: ' || TO_CHAR(sa));
dbms_output.put_line('Chained Percentage: ' || TO_CHAR(cp));
END;
/
Space Used: 1055374677
Space Allocated: 1073741824
Chained Percentage: 0
Related
I (using Oracle 12c, PL/SQL) need to update an existing table TABLE1 based on information stored in a table MAP. In a simplified version, MAP looks like this:
COLUMN_NAME
MODIFY
COLUMN1
N
COLUMN2
Y
COLUMN3
N
...
...
COLUMNn
Y
COLUMN1 to COLUMNn are column names in TABLE1 (but there are more columns, not just these). Now I need to update a column in TABLE1 if MODIFY in table MAP contains a 'Y' for that columns' name. There are other row conditions, so what I would need would be UPDATE statements of the form
UPDATE TABLE1
SET COLUMNi = value_i
WHERE OTHER_COLUMN = 'xyz_i';
where COLUMNi runs through all the columns of TABLE1 which are marked with MODIFY = 'Y' in MAP. value_i and xyz_i also depend on information stored in MAP (not displayed in the example).
The table MAP is not static but changes, so I do not know in advance which columns to update. What I did so far is to generate the UPDATE-statements I need in a query from MAP, i.e.
SELECT <Text of UPDATE-STATEMENT using row information from MAP> AS SQL_STMT
FROM MAP
WHERE MODIFY = 'Y';
Now I would like to execute these statements (possibly hundreds of rows). Of course I could just copy the contents of the query into code and execute, but is there a way to do this automatically, e.g. using EXECUTE IMMEDIATE? It could be something like
BEGIN
EXECUTE IMMEDIATE SQL_STMT USING 'xyz_i';
END;
only that SQL_STMT should run through all the rows of the previous query (and 'xyz_i' varies with the row as well). Any hints how to achieve this or how one should approach the task in general?
EDIT: As response to the comments, a bit more background how this problem emerges. I receive an empty n x m Matrix (empty except row and column names, think of them as first row and first column) quarterly and need to populate the empty fields from another process.
The structure of the initial matrix changes, i.e. there may be new/deleted columns/rows and existing columns/rows may change their position in the matrix. What I need to do is to take the old version of the matrix, where I already have filled the empty spaces, and translate this into the new version. Then, the populating process merely looks if entries have changed and if so, alters them.
The situation from the question arises after I have translated the old version into the new one, before doing the delta. The new matrix, populated with the old information, is TABLE1. The delta process, over which I have no control, gives me column names and information to be entered into the cells of the matrix (this is table MAP). So I need to find the column in the matrix labeled by the delta process and then to change values in rows (which ones is specified via other information provided by the delta process)
Dynamic SQL it is; here's an example, see if it helps.
This is a table whose contents should be modified:
SQL> select * from test order by id;
ID NAME SALARY
---------- ---------- ----------
1 Little 100
2 200
3 Foot 0
4 0
This is the map table:
SQL> select * from map;
COLUMN CB_MODIFY VALUE WHERE_CLAUSE
------ ---------- ----- -------------
NAME Y Scott where id <= 3
SALARY N 1000 where 1 = 1
Procedure loops through all columns that are set to be modified, composes the dynamic update statement and executes it:
SQL> declare
2 l_str varchar2(1000);
3 begin
4 for cur_r in (select m.column_name, m.value, m.where_clause
5 from map m
6 where m.cb_modify = 'Y'
7 )
8 loop
9 l_str := 'update test set ' ||
10 cur_r.column_name || ' = ' || chr(39) || cur_r.value || chr(39) || ' ' ||
11 cur_r.where_clause;
12 execute immediate l_str;
13 end loop;
14 end;
15 /
PL/SQL procedure successfully completed.
Result:
SQL> select * from test order by id;
ID NAME SALARY
---------- ---------- ----------
1 Scott 100
2 Scott 200
3 Scott 0
4 0
SQL>
Is there a hint to generate execution plan ignoring the existing one from the shared pool?
There is not a hint to create an execution plan that ignores plans in the shared pool. A more common way of phrasing this question is: how do I get Oracle to always perform a hard parse?
There are a few weird situations where this behavior is required. It would be helpful to fully explain your reason for needing this, as the solution varies depending why you need it.
Strange performance problem. Oracle performs some dynamic re-optimization of SQL statements after the first run, like adaptive cursor sharing and cardinality feedback. In the rare case when those features backfire you might want to disable them.
Dynamic query. You have a dynamic query that used Oracle data cartridge to fetch data in the parse step, but Oracle won't execute the parse step because the query looks static to Oracle.
Misunderstanding. Something has gone wrong and this is an XY problem.
Solutions
The simplest way to solve this problem are by using Thorsten Kettner's solution of changing the query each time.
If that's not an option, the second simplest solution is to flush the query from the shared pool, like this:
--This only works one node at a time.
begin
for statements in
(
select distinct address, hash_value
from gv$sql
where sql_id = '33t9pk44udr4x'
order by 1,2
) loop
sys.dbms_shared_pool.purge(statements.address||','||statements.hash_value, 'C');
end loop;
end;
/
If you have no control over the SQL, and need to fix the problem using a side-effect style solution, Jonathan Lewis and Randolf Geist have a solution using Virtual Private Database, that adds a unique predicate to each SQL statement on a specific table. You asked for something weird, here's a weird solution. Buckle up.
-- Create a random predicate for each query on a specific table.
create table hard_parse_test_rand as
select * from all_objects
where rownum <= 1000;
begin
dbms_stats.gather_table_stats(null, 'hard_parse_test_rand');
end;
/
create or replace package pkg_rls_force_hard_parse_rand is
function force_hard_parse (in_schema varchar2, in_object varchar2) return varchar2;
end pkg_rls_force_hard_parse_rand;
/
create or replace package body pkg_rls_force_hard_parse_rand is
function force_hard_parse (in_schema varchar2, in_object varchar2) return varchar2
is
s_predicate varchar2(100);
n_random pls_integer;
begin
n_random := round(dbms_random.value(1, 1000000));
-- s_predicate := '1 = 1';
s_predicate := to_char(n_random, 'TM') || ' = ' || to_char(n_random, 'TM');
-- s_predicate := 'object_type = ''TABLE''';
return s_predicate;
end force_hard_parse;
end pkg_rls_force_hard_parse_rand;
/
begin
DBMS_RLS.ADD_POLICY (USER, 'hard_parse_test_rand', 'hard_parse_policy', USER, 'pkg_rls_force_hard_parse_rand.force_hard_parse', 'select');
end;
/
alter system flush shared_pool;
You can see the hard-parsing in action by running the same query multiple times:
select * from hard_parse_test_rand;
select * from hard_parse_test_rand;
select * from hard_parse_test_rand;
select * from hard_parse_test_rand;
Now there are three entries in GV$SQL for each execution. There's some odd behavior in Virtual Private Database that parses the query multiple times, even though the final text looks the same.
select *
from gv$sql
where sql_text like '%hard_parse_test_rand%'
and sql_text not like '%quine%'
order by 1;
I think there is no hint indicating that Oracle shall find a new execution plan everytime it runs the query.
This is something we'd want for select * from mytable where is_active = :active, with is_active being 1 for very few rows and 0 for maybe billions of other rows. We'd want an index access for :active = 1 and a full table scan for :active = 0 then. Two different plans.
As far as I know, Oracle uses bind variable peeking in later versions, so with a look at the statistics it really comes up with different execution plans for different bind varibale content. But in older versions it did not, and thus we'd want some hint saying "make a new plan" there.
Oracle only re-used an execution plan for exactly the same query. It sufficed to add a mere blank to get a new plan. Hence a solution might be to generate the query everytime you want to run it with a random number included in a comment:
select /* 1234567 */ * from mytable where is_active = :active;
Or just don't use bind variables, if this is the problem you want to address:
select * from mytable where is_active = 0;
select * from mytable where is_active = 1;
I have 2 delete statements that are taking a long time to complete. There are several indexes on the columns in where clause.
What is a duplicate?
If 2 or more records have same values in columns id,cid,type,trefid,ordrefid,amount and paydt then there are duplicates.
The DELETEs delete about 1 million record.
Can they be re-written in any way to make it quicker.
DELETE FROM TABLE1 A WHERE loaddt < (
SELECT max(loaddt) FROM TABLE1 B
WHERE
a.id=b.id and
a.cid=b.cid and
NVL(a.type,'-99999') = NVL(b.type,'-99999') and
NVL(a.trefid,'-99999')=NVL(b.trefid,'-99999') and
NVL(a.ordrefid,'-99999')= NVL(b.ordrefid,'-99999') and
NVL(a.amount,'-99999')=NVL(b.amount,'-99999') and
NVL(a.paydt,TO_DATE('9999-12-31','YYYY-MM-DD'))=NVL(b.paydt,TO_DATE('9999-12-31','YYYY-MM-DD'))
);
COMMIT;
DELETE FROM TABLE1 a where rowid > (
Select min(rowid) from TABLE1 b
WHERE
a.id=b.id and
a.cid=b.cid and
NVL(a.type,'-99999') = NVL(b.type,'-99999') and
NVL(a.trefid,'-99999')=NVL(b.trefid,'-99999') and
NVL(a.ordrefid,'-99999')= NVL(b.ordrefid,'-99999') and
NVL(a.amount,'-99999')=NVL(b.amount,'-99999') and
NVL(a.paydt,TO_DATE('9999-12-31','YYYY-MM-DD'))=NVL(b.paydt,TO_DATE('9999-12-31','YYYY-MM-DD'))
);
commit;
Explain Plan:
DELETE TABLE1
HASH JOIN 1296491
Access Predicates
AND
A.ID=ITEM_1
A.CID=ITEM_2
ITEM_3=NVL(TYPE,'-99999')
ITEM_4=NVL(TREFID,'-99999')
ITEM_5=NVL(ORDREFID,'-99999')
ITEM_6=NVL(AMOUNT,(-99999))
ITEM_7=NVL(PAYDT,TO_DATE(' 9999-12-31 00:00:00', 'syyyy-mm-dd hh24:mi:ss'))
Filter Predicates
LOADDT<MAX(LOADDT)
TABLE ACCESS TABLE1 FULL 267904
VIEW VW_SQ_1 690385
SORT GROUP BY 690385
TABLE ACCESS TABLE1 FULL 267904
How large is the table? If count of deleted rows is up to 12% then you may think about index.
Could you somehow partition your table - like week by week and then scan only actual week?
Maybe this could be more effecient. When you're using aggregate function, then oracle must walk through all relevant rows (in your case fullscan), but when you use exists it stops when the first occurence is found. (and of course the query would be much faster, when there was one function-based(because of NVL) index on all columns in where clause)
DELETE FROM TABLE1 A
WHERE exists (
SELECT 1
FROM TABLE1 B
WHERE
A.loaddt != b.loaddt
a.id=b.id and
a.cid=b.cid and
NVL(a.type,'-99999') = NVL(b.type,'-99999') and
NVL(a.trefid,'-99999')=NVL(b.trefid,'-99999') and
NVL(a.ordrefid,'-99999')= NVL(b.ordrefid,'-99999') and
NVL(a.amount,'-99999')=NVL(b.amount,'-99999') and
NVL(a.paydt,TO_DATE('9999-12-31','YYYY-MM-DD'))=NVL(b.paydt,TO_DATE('9999-12-31','YYYY-MM-DD'))
);
Although some may disagree, I am a proponent of running large, long running deletes procedurally. In my view it is much easier to control and track progress (and your DBA will like you better ;-) Also, not sure why you need to join table1 to itself to identify duplicates (and I'd be curious if you ever run into snapshot too old issues with your current approach). You also shouldn't need multiple delete statements, all duplicates should be handled in one process. Finally, you should check WHY you're constantly re-introducing duplicates each week, and perhaps change the load process (maybe doing a merge/upsert rather than all inserts).
That said, you might try something like:
-- first create mat view to find all duplicates
create materialized view my_dups_mv
tablespace my_tablespace
build immediate
refresh complete on demand
as
select id,cid,type,trefid,ordrefid,amount,paydt, count(1) as cnt
from table1
group by id,cid,type,trefid,ordrefid,amount,paydt
having count(1) > 1;
-- dedup data (or put into procedure and schedule along with mat view refresh above)
declare
-- make sure my_dups_mv is refreshed first
cursor dup_cur is
select * from my_dups_mv;
type duprec_t is record(row_id rowid);
duprec duprec_t;
type duptab_t is table of duprec_t index by pls_integer;
duptab duptab_t;
l_ctr pls_integer := 0;
l_dupcnt pls_integer := 0;
begin
for rec in dup_cur
loop
l_ctr := l_ctr + 1;
-- assuming needed indexes exist
select rowid
bulk collect into duptab
from table1
where id = rec.id
and cid = rec.cid
and type = rec.type
and trefid = rec.trefid
and ordrefid = rec.ordrefid
and amount = rec.amount
and paydt = rec.paydt
-- order by whatever makes sense to make the "keeper" float to top
order by loaddt desc
;
for i in 2 .. duptab.count
loop
l_dupcnt := l_dupcnt + 1;
delete from table1 where rowid = duptab(i).row_id;
end loop;
if (mod(l_ctr, 10000) = 0) then
-- log to log table here (calling autonomous procedure you'll need to implement)
insert_logtable('Table1 deletes', 'Commit reached, deleted ' || l_dupcnt || ' rows');
commit;
end if;
end loop;
commit;
end;
Check your log table for progress status.
1. Parallel
alter session enable parallel dml;
DELETE /*+ PARALLEL */ FROM TABLE1 A WHERE loaddt < (
...
Assuming you have Enterprise Edition, a sane server configuration, and you are on 11g. If you're not on 11g, the parallel syntax is slightly different.
2. Reduce memory requirements
The plan shows a hash join, which is probably a good thing. But without any useful filters, Oracle has to hash the entire table. (Tbone's query, that only use a GROUP BY, looks nicer and may run faster. But it will also probably run into the same problem trying to sort or hash the entire table.)
If the hash can't fit in memory it must be written to disk, which can be very slow. Since you run this query every week, only one of the tables needs to look at all the rows. Depending on exactly when it runs, you can add something like this to the end of the query: ) where b.loaddt >= sysdate - 14. This may significantly reduce the amount of writing to temporary tablespace. And it may also reduce read IO if you use some partitioning strategy like jakub.petr suggested.
3. Active Report
If you want to know exactly what your query is doing, run the Active Report:
select dbms_sqltune.report_sql_monitor(sql_id => 'YOUR_SQL_ID_HERE', type => 'active')
from dual;
(Save the output to an .html file and open it with a browser.)
There are plenty of stored procedure on 10g platform . ( Almost 500 SPs)
Each SP might have loop, fetch and etc.
I'd like to ask you if there is a cool method to control all the SPs which are currently running on 10g, and guarantee that it works on 11 g.
I have a development server 1 which is 10 g and the other development server is 11 g.
I can use both of them to testify the propose above.
For instance I know that on 10 g if you use loop, and during the loop the update statements do not affect the loop data but 11g.
There might be more cases that I have to consider. Please tell me if you have any brillant idea , otherwise I will check them up one by one manually and it is a lot of time and human control might be weak sometimes.
important note: It is said that if you select some data from a table or tables, and if you use it in a loop, then during de loop, if you update and commit between loop case, it affects the selected data in cursor.(#11g) But this did not happen #10g version. Please correct me if you heard something like that.
The Example Case;
CREATE TABLE vty_musteri(
musterino NUMBER NOT NULL,
subeadi VARCHAR2(61),
kayitzamani VARCHAR2(20)
);
INSERT INTO vty_musteri (musterino, subeadi, kayitzamani )
VALUES (12345, 'AMSTERDAM', '05/30/2012 15:11:13');
COMMIT;
CREATE UNIQUE INDEX vty_musteri_idx ON vty_musteri (musterino);
SELECT * FROM vty_musteri;
CREATE OR REPLACE PROCEDURE krd_upd_silseomusteri_sp(RC1 in out SYS_REFCURSOR) AS
v_musterino NUMBER := 12345;
BEGIN
OPEN RC1 FOR
SELECT m.musterino, m.subeadi, m.kayitzamani
FROM vty_musteri m
WHERE m.musterino = v_musterino;
update vty_musteri
set subeadi = 'PORTO',
kayitzamani = (SELECT TO_CHAR(SYSDATE, 'MM/DD/YYYY HH24:MI:SS')
FROM dual)
where musterino = v_musterino;
COMMIT;
After all run this test on PLSQL:
DECLARE
--test
vRecTip SYS_REFCURSOR;
TYPE vRecTipK IS RECORD(
musterino NUMBER,
subeadi VARCHAR2(61),
kayitzamani VARCHAR2(20)
);
v_SeoTip vRecTipK;
BEGIN
krd_upd_silseomusteri_sp(rc1 => vRecTip);
IF vRecTip%ISOPEN THEN
LOOP
FETCH vRecTip
INTO v_SeoTip;
EXIT WHEN vRecTip%NOTFOUND;
dbms_output.put_line('The Value : ' || v_SeoTip.musterino || ' - ' || v_SeoTip.subeadi || ' - ' || v_SeoTip.kayitzamani);
END LOOP;
END IF;
COMMIT;
END;
END;
If you run this on 10g you will see AMSTERDAM, but on 11G, it is PORTO.
To fix it; I put a hint in the sp like the following:
SELECT /*+ full(m)*/ m.musterino, m.subeadi, m.kayitzamani
Isn't it weird? any alternative sugesstion to get AMSTERDAM ?
One thing we stumbled upon during a migration were queries that weren't supposed to work on 10.x (but did anyway) did no longer work on 11.x
This happens if you have ambigous column references in your query.
Something like this:
SELECT name,
f.some_col,
b.other_col
FROM foo f,
JOIN bar b ON f.id = b.fid
If the column name exists in both tables, 10.x would run the statement - which was a bug.
This bug (BugID: 6760937) was fixed and makes the statement (rightfully) fail in 11.x
Basic PLSQL structures should work exactly the same. Some pitfalls are listed here:
http://www.help2ora.com/index.php/2011/08/04/be-careful-when-migrating-difference-between-oracle-10g-and-11g/
To fix it; I put a hint in the sp like the following:
SELECT /+ full(m)/ m.musterino, m.subeadi, m.kayitzamani
Recently I have done migration to Oracle 11g. Faced few unprecedented issues. I have written a blog post on this. Have a look http://learncodewrite.blogspot.in/2017/04/migrating-to-oracle-11g-from-oracle-10g.html?m=1.
We just converted our sql server stored procedures to oracle procedures. Sql Server SP's were highly dependent on session tables (INSERT INTO #table1...) these tables got converted as global temporary tables in oracle. We ended up with aroun 500 GTT's for our 400 SP's
Now we are finding out that working with GTT's in oracle is considered a last option because of performance and other issues.
what other alternatives are there? Collections? Cursors?
Our typical use of GTT's is like so:
Insert into GTT
INSERT INTO some_gtt_1
(column_a,
column_b,
column_c)
(SELECT someA,
someB,
someC
FROM TABLE_A
WHERE condition_1 = 'YN756'
AND type_cd = 'P'
AND TO_NUMBER(TO_CHAR(m_date, 'MM')) = '12'
AND (lname LIKE (v_LnameUpper || '%') OR
lname LIKE (v_searchLnameLower || '%'))
AND (e_flag = 'Y' OR
it_flag = 'Y' OR
fit_flag = 'Y'));
Update the GTT
UPDATE some_gtt_1 a
SET column_a = (SELECT b.data_a FROM some_table_b b
WHERE a.column_b = b.data_b AND a.column_c = 'C')
WHERE column_a IS NULL OR column_a = ' ';
and later on get the data out of the GTT. These are just sample queries, in actuality the queries are really complext with lot of joins and subqueries.
I have a three part question:
Can someone show how to transform
the above sample queries to
collections and/or cursors?
Since
with GTT's you can work natively
with SQL...why go away from the
GTTs? are they really that bad.
What should be the guidelines on
When to use and When to avoid GTT's
Let's answer the second question first:
"why go away from the GTTs? are they
really that bad."
A couple of days ago I was knocking up a proof of concept which loaded a largish XML file (~18MB) into an XMLType. Because I didn't want to store the XMLType permanently I tried loading it into a PL/SQL variable (session memory) and a temporary table. Loading it into a temporary table took five times as long as loading it into an XMLType variable (5 seconds compared to 1 second). The difference is because temporary tables are not memory structures: they are written to disk (specifically your nominated temporary tablespace).
If you want to cache a lot of data then storing it in memory will stress the PGA, which is not good if you have lots of sessions. So it's a trade-off between RAM and time.
To the first question:
"Can someone show how to transform the
above sample queries to collections
and/or cursors?"
The queries you post can be merged into a single statement:
SELECT case when a.column_a IS NULL OR a.column_a = ' '
then b.data_a
else column_a end AS someA,
a.someB,
a.someC
FROM TABLE_A a
left outer join TABLE_B b
on ( a.column_b = b.data_b AND a.column_c = 'C' )
WHERE condition_1 = 'YN756'
AND type_cd = 'P'
AND TO_NUMBER(TO_CHAR(m_date, 'MM')) = '12'
AND (lname LIKE (v_LnameUpper || '%') OR
lname LIKE (v_searchLnameLower || '%'))
AND (e_flag = 'Y' OR
it_flag = 'Y' OR
fit_flag = 'Y'));
(I have simply transposed your logic but that case() statement could be replaced with a neater nvl2(trim(a.column_a), a.column_a, b.data_a) ).
I know you say your queries are more complicated but your first port of call should be to consider rewriting them. I know how seductive it is to break a gnarly query into lots of baby SQLs stitched together with PL/SQL but pure SQL is way more efficient.
To use a collection it is best to define the types in SQL, because it gives us the flexibility to use them in SQL statements as well as PL/SQL.
create or replace type tab_a_row as object
(col_a number
, col_b varchar2(23)
, col_c date);
/
create or replace type tab_a_nt as table of tab_a_row;
/
Here's a sample function, which returns a result set:
create or replace function get_table_a
(p_arg in number)
return sys_refcursor
is
tab_a_recs tab_a_nt;
rv sys_refcursor;
begin
select tab_a_row(col_a, col_b, col_c)
bulk collect into tab_a_recs
from table_a
where col_a = p_arg;
for i in tab_a_recs.first()..tab_a_recs.last()
loop
if tab_a_recs(i).col_b is null
then
tab_a_recs(i).col_b := 'something';
end if;
end loop;
open rv for select * from table(tab_a_recs);
return rv;
end;
/
And here it is in action:
SQL> select * from table_a
2 /
COL_A COL_B COL_C
---------- ----------------------- ---------
1 whatever 13-JUN-10
1 12-JUN-10
SQL> var rc refcursor
SQL> exec :rc := get_table_a(1)
PL/SQL procedure successfully completed.
SQL> print rc
COL_A COL_B COL_C
---------- ----------------------- ---------
1 whatever 13-JUN-10
1 something 12-JUN-10
SQL>
In the function it is necessary to instantiate the type with the columns, in order to avoid the ORA-00947 exception. This is not necessary when populating a PL/SQL table type:
SQL> create or replace procedure pop_table_a
2 (p_arg in number)
3 is
4 type table_a_nt is table of table_a%rowtype;
5 tab_a_recs table_a_nt;
6 begin
7 select *
8 bulk collect into tab_a_recs
9 from table_a
10 where col_a = p_arg;
11 end;
12 /
Procedure created.
SQL>
Finally, guidelines
"What should be the guidelines on When
to use and When to avoid GTT's"
Global temp tables are very good when we need share cached data between different program units in the same session. For instance if we have a generic report structure generated by a single function feeding off a GTT which is populated by one of several procedures. (Although even that could also be implemented with dynamic ref cursors ...)
Global temporary tables are also good if we have a lot of intermediate processing which is just too complicated to be solved with a single SQL query. Especially if that processing must be applied to subsets of the retrieved rows.
But in general the presumption should be that we don't need to use a temporary table. So
Do it in SQL unless it is too hard it which case ...
... Do it in PL/SQL variables (usually collections) unless it takes too much memory it which case ...
... Do it with a Global Temporary Table
Generally I'd use a PL/SQL collection for storing small volumes of data (maybe a thousand rows). If the data volumes were much larger, I'd use a GTT so that they don't overload the process memory.
So I might select a few hundred rows from the database into a PL/SQL collection, then loop through them to do some calculation/delete a few or whatever, then insert that collection into another table.
If I was dealing with hundreds of thousands of rows, I would try to push as much of the 'heavy lifting' processing into large SQL statements. That may or may not require GTT.
You can use SQL level collection objects as something that translates quite easily between SQL and PL/SQL
create type typ_car is object (make varchar2(10), model varchar2(20), year number(4));
/
create type typ_coll_car is table of typ_car;
/
select * from table (typ_coll_car(typ_car('a','b',1999), typ_car('A','Z',2000)));
MAKE MODEL YEAR
---------- -------------------- ---------------
a b 1,999.00
A Z 2,000.00
declare
v_car1 typ_car := typ_car('a','b',1999);
v_car2 typ_car := typ_car('A','Z',2000);
t_car typ_coll_car := typ_coll_car();
begin
t_car := typ_coll_car(v_car1, v_car2);
FOR i in (SELECT * from table(t_car)) LOOP
dbms_output.put_line(i.year);
END LOOP;
end;
/