I want to know which is the most efficient insert method among the followiing
1)using the hint append
or
2)bulk collect. Preferably I'd like to know which method of inserting for say 2-5 millions rows is the best.
I've created a materialized view log on table with the following statement:
CREATE MATERIALIZED VIEW LOG ON table_a
WITH ROWID, SEQUENCE (column_a, column_b, column_c)
INCLUDING NEW VALUES;
The insert is done with the following statement:
INSERT
/*+ APPEND PARALLEL("table_a") */
INTO
"table_a"
("column_a",
"column_b",
"column_c",
"column_d_sum")
(select
column_a",
"column_b",
"column_c",
"column_d_sum"
from table_B)
But the Log is empty when the insert is finished. When I insert rows without the APPEND hint, rows are created in the log table. So, doesn't the log record bulk loads?
i used sql loader to import data from csv file to my db.but every time the columns places are changed.o i need dynamic way to insert data into correct column in the table.
in csv file contains column name and i insert this data to temp table, after that i want to read data over column name.also i read the column names from (All_Tab_Columns) to make combination of column name between temp table and All_Tab_Columns table to insert data to right place...
I made two runs for bulk insertion
In first run, 16,36,897 were inserted successfully in around 38 seconds.But in second run, 54,62,952 records had to be inserted, but process failed after 708 seconds with following error :
Error report:
ORA-04030: out of process memory when trying to allocate 980248 bytes (PLS non-lib hp,DARWIN)
ORA-06512: at line 21
04030. 00000 - "out of process memory when trying to allocate %s bytes (%s,%s)"
*Cause: Operating system process private memory has been exhausted
*Action:
Here is my code snippet :
.......
FORALL i in products_tab.first .. products_tab.last
INSERT INTO tab1 VALUES products_tab(i);
COMMIT;
.........
I think that there should not have been any problem in getting it completed successfully.
We have requirement to create INSERT SCRIPT from the table having thousands of records and load that into flat file. if there are any better option other than using UTL_FILE package which process record by record.
View 5 Replies View RelatedWe are doing a bulk select and insert (10,000 rows processed in each transaction). If one record fails, the entire transaction is rolled out. We need to fix this and re-run. the process is repeated unless all errors are fixed.
How to capture all errors in a single run ?
why bulk insert is not possible in a table which has index?
View 9 Replies View RelatedI have a table emp as mentioned below:
SELECT * FROM EMP_TEST;
EMP_TEST
-----------------------------------------------------
ENO EFIRSTNAME ESECONDNAME DEPTNO
1 JOHN PAI 10
2 ABC DEF 20
3 EFG GHI 30
Now the primary key in this above table is pk_emp_test_eno on eno.
I have a requirement where i need to dump some dummy data (600000000 numbers of data ) into the emp_test based on these existing data without disabling the constraints (maintaining unique constraint for each record). And while inserting i want to commit after every 1000 insertion.
in bulk inserting dummy datas into the table as it is taking much more time to insert into the same.
I used bulk collect and for all statements to select and insert the data in temp table.The select SQl is returning one row. But its not inserting this row into temp table.Its not throwing any exceptions. Used ref cursor because the select statement is going for every cursor.
here modified the code and provided only one cursor.
Create Or Replace Procedure Sales_Hist_Update_Bkp Is
Type Type_Name Is Record(
Sku_Item_Key Ordm_Int.Dwi_Rtl_Sls_Retrn_Line_Bkp.Sku_Item_Key%Type,
Locationno Ordm_Int.Dwi_Rtl_Sls_Retrn_Line_Bkp.Locationno%Type,
Bsns_Unit_Key Ordm_Int.Dwi_Rtl_Sls_Retrn_Line_Bkp.Bsns_Unit_Key%Type,
Act_Item_Cost_Amt Ordm_Int.Dwi_Rtl_Sls_Retrn_Line_Bkp.Item_Cost_Amt%Type,
Act_Rglr_Unit_Price_Amt Ordm_Int.Dwi_Rtl_Sls_Retrn_Line_Bkp.Rglr_Unit_Price_Amt%Type,
[code]...
Is there any defined record count range for the following ways of bulk insert :
INSERT INTO ABCTEMP SELECT * FROM DEFTEMP;
OR
through a cursor, bulk fetch and bulk insert under a loop.
SQL> declare
2 TYPE id_collection is TABLE of number(6);
3 TYPE ename_collection is TABLE of varchar2(20);
4 id ID_COLLECTION;
5 ename ENAME_COLLECTION;
6 cursor c is select empid,name from Nemp;
7 begin
8 open c;
9 loop
[code]....
Here sub_Nemp is my new table in which i have to insert the values from Nemp old table.Both tables are same like below:-
SQL> create table sub_Nemp(empid number(6),name varchar2(20));
I'm unable to find this error...
I have an array of C structs say
struct S
{
int a, long b, double c;
};
S s[100];
Further suppose I have an Oracle table T like this:
create table T
(
a number(10),
b number(10),
c float
);
I want to bulk insert all 100 instances of S from a client application into T. I've seen code that does this for *one* field or column. The code defines a stored procedure which accepts a single argument which is a TABLE and then does a FORALL ... insert. The client application passes in the array of data.
What I need is N columns. In my example above struct S has N=3 fields which conform to the N=3 columns in T. In reality my N will be 50+. I am trying to avoid creating stored procedures which will take the 50 or so arguments it will eventually need.
So does my stored procedure need to accept N TABLE arguments? Or can I cajole OCI/OTL/ODBC and PL/SQL so that the stored procedure can take an array of rows which the type of row conforms to T by defining a record or something? That is, do I need:
Option 1: // declares one type and one argument each for N cols
create or replace procedure insert_S(
a_array IN A_TABLE, -- type A_TABLE is TABLE of number;
b_array IN B_TABLE, -- type B_TABLE is TABLE of number;
c_array IN C_TABLE) -- type C_TABLE is ...
begin ... end
Option 2: // this somehow accepts an array compatible with T
// if I could get a OCI/OCCI/OTL/ODBC application
// to send this data, this procedure would have
// only one argument
create or replace procedure insert_S(
row_array IN ?????????? type -- some sort of array of rows
)
begin ... end
Or should I pass the whole memory chunk of data in as an image or varchar array -- basically an opaque block of data -- and then internally decypher/decode the memory block inside the stored procedure as discussed on [URL].
best way to pass an array of N C-structs of M fields to a stored procedure for insertion into a table with M compatible columns? One TABLE per column? with an array of a custom type compatible with a row in T? As glob of data? Another option is to populate some host variables ... but, again, I'd need N host variables.
I am facing a problem in bulk insert using SELECT statement.My sql statement is like below.
strQuery :='INSERT INTO TAB3
(SELECT t1.c1,t2.c2
FROM TAB1 t1, TAB2 t2
WHERE t1.c1 = t2.c1
AND t1.c3 between 10 and 15 AND)' ....... some other conditions.
EXECUTE IMMEDIATE strQuery...These SQL statements are inside a procedure. And this procedure is called from C#.The number of rows returned by the "SELECT" query is 70.
On the very first time call of this procedure, the number rows inserted using strQuery is 70. But in the next time call (in the same transaction) of the procedure, the number rows inserted is only 50.And further if we are repeating calling this procedure, it will insert sometimes 70 or 50 etc. It is showing some inconsistency.On my initial analysis it is found that, the default optimizer is "ALL_ROWS". When i changed the optimizer mode to "rule", this issue is not coming.I am using Oracle 10g R2 version.
Using the Bulk collect for insert into table,it's raising the below error.
ORA-00600: internal error code, arguments: [25027], [130], [1], [], [], [], [], [], [], [], [], []
I am working on oracle 11g...I have one normal insert proc
CREATE OR REPLACE PROCEDURE test2
AS
BEGIN
INSERT INTO first_table
(citiversion, financialcollectionid,
dataitemid, dataitemvalue,
[code]....
I am processing 1 lakh rows.tell me the reason why bulk collect is taking more time. ? According to my knowledge it should take less time. do i need to check any parameter?
Table contains 10k records,we are going to insert data into another table with FORALL bulk collect limit 1000. if i use 10000 ,it's completed fast compared to 1000 limit.Can u tell me which one is better Limit.
View 4 Replies View RelatedPerhaps this is a common request : I have 2 tables:
Table A
-------
ID Value
1 a
2 b
3 c
Table B
-------
ID AnotherValue
1 x
2 y
I am hoping to append a column from Table B to Table A based on a simple sql join (e.g:
Table A
ID Value AnotherValue
1 a x
2 b y
3 c (null)
)
I would rather stay away from the standard update statement since it takes far to long and I'd prefer not to use create table as I don't want to duplicate any data...is this possible to do ? (e.g: just insert the columns into this table ?) - or if it's possible the performance overhead just wouldn't make it worth it ?
Declare
Cursor c1...;
Cursor c2...;
begin
open c1 ;
fecth c1 bulk collect into v1;
close c1;
[Code]...
Is there any way by which if condition gets true then v1 gets appended rather than being overwritten?
declare
type lst_deptno is table of dept.deptno%type index by binary_integer;
type lst_deptno_emp is table of emp.deptno%type index by binary_integer;
v_deptno lst_deptno;
v_deptno_emp lst_deptno_emp;
cursor c1 is select deptno from dept;
[Code]...
if the same name repeating it should to append with _1 and _2 until same name reached.
select 'fname' name from dual
union all
select 'lname' name from dual
union all
select 'email' name from dual
union all
select 'fname' name from dual
union all
select 'fname' name from dualmy output should be like below...
fname
lname
email
fname_1
fname_2
I am just looking to my control file that I have. I have managed to successfully get my control file to load data into the required tables but just for my learning I was trying to amend the when clause with an and condition but keep getting an error.
My code:
OPTIONS (DIRECT=TRUE, ERRORS=0, SKIP=1)
load data
infile 'A:/My Files/Feeds/20130211/data_summary_1_20131214.txt' "STR '
'"
APPEND
into table I_DATA_1
WHEN (RID <> 'RID')
fields terminated by "~|~" optionally enclosed by '"'
TRAILING NULLCOLS
(
RID,
S_TIMESTAMP timestamp 'DD-MON-RR HH:MI:SS.FF3 PM'
)
the above code works perfectly but if i change the following line:
WHEN (RID <> 'RID' AND 'S_TIMESTAMP' IS NOT NULL)than this gives me an error.
I have a requirement when I need to append and show the contents of a CLOB (rich text) column into the Open office report.
The use case is as below.
User(s) can enter the details into a CLOB column.
This data entered by each user need to be concatenated onto another CLOB column which will hold all the history of changes. Mainly i wanted to perform a concatenation of these two columns.
|
| User Entered | Stored |
| Data Stored in Clob Column 1| Data Stored in Clob Column 2 |
| v_clob_1 | v_clob_target |
-------------------------------------------------------------------------------
|"Text 1" and a image |"Text 1" and a image |
-------------------------------------------------------------------------------
Any method to achieve this case. I had tried the following method to achieve this. But no success.
v_merge clob;
v_clob_target clob;
v_clob_1 clob;
1)
DBMS_LOB.APPEND(v_merge, v_clob_target);
DBMS_LOB.APPEND(v_merge, v_clob_1);
2)
DBMS_LOB.APPEND(v_clob_target, v_clob_1);
3)
v_merge:= v_clob_target || ''|| v_clob_1;
v_clob_target := v_merge;
Have got ddl of index,but it is not append ';' end of each index,how to append ';' by auto.
Select
Dbms_Metadata.Get_Dependent_Ddl(Object_Type => 'INDEX',
Base_Object_Name => 'TB_HXL_USER')
From Dual
result:
[Code].....
Version: 11.2.0.3
Platform : RHEL 5.8 (But I am looking for platform independant solution)
I want to append the timestamp to spooled log file name in SQL*Plus.The spooled log filename should look like
WMS_APP_23-March-2013.logI tried the following 3 methods found in the google. But none of them worked !
I tried this
col sysdt noprint new_value sysdt_var
SELECT TO_CHAR(SYSDATE, 'yyyymmdd_hh24miss') sysdt FROM DUAL;
spool run_filename_&sysdt_var.Logas suggested in
[URL]
and this
spool filename with timestamp
col sysdt noprint new_value sysdt
SELECT TO_CHAR(SYSDATE, 'yyyymmdd_hh24miss') sysdt FROM DUAL;
spool run_filename_&sysdt..Logas suggested in
[URL]
and this
column tm new_value file_time noprint
select to_char(sysdate, 'YYYYMMDD') tm from dual ;
prompt &file_time
spool logfile_id&file_time..logas suggested in
Creating a spool file with date/time appended to file name
None of the above worked in RHEL or MS DOS.
I have a table of below structure:
CREATE TABLE Emp_addrs
(
EMP_ID NUMBER(15) NOT NULL,
ADDRESS_ID NUMBER(15) NOT NULL,
SITE_USE_ID NUMBER(15) NOT NULL,
SITE_USE_STATUS VARCHAR2(1 BYTE) NOT NULL,
SITE_USE_CODE VARCHAR2(30 BYTE) NOT NULL)
Insert Script code :
insert into Emp_addrs values ( '1207' , '1846', '2342','A');
insert into Emp_addrs values ( '1207' , '1846', '2343','I');
insert into Emp_addrs values ( '61618' , '165200', '261449','A');
[Code]...
A combination of emp_id & address_id can have multiple site_use_id's. I want to select the max(site_use_id) where site_use_status ='A'.
Now Site_use_status can have either = 'I' or 'A' value.
For a combination of emp_id and address_id , there could be cases when there is no record with site_use_status ='A'. In such cases I need to select the max(site_use_id) (and obviously site_use_status ='I').
Just to clear my requirements, out of the above records I want the following records:
'1207' , '1846', '2342','A'
'61618' , '165200', '261449','A'
'1003' , '1004', '1007','A'
'1005' , '1010', '2002','I'
'2005' , '2010', '3002','A'
i am using one query but not getting correct minutes.
here is my query:
v_Interval:= to_timestamp(v_temphrs,'HH24:MI:SS')-to_timestamp(v_outpunch1,'HH24:MI:SS');
v_TotalHrsMin1 := extract(hour from v_interval) * 60 + extract(minute from v_interval);
here v_interval datatype is "interval day to second" and v_temphrs datatype is varchar2 and value is : 12:00:00 and v_outpunch1 datatype is varchar2 and value is: 06:10:00
and v_totalHrsMin1 datatype is number.
here i should get value 370.
but i am getting value 350.
CREATE OR REPLACE VIEW V_CATALOGUER_REPORT
(COUNT, EVENT_USER, TO_DATE, SHORT_NAME)
AS
SELECT COUNT (DISTINCT A.PART_REF),A.EVENT_USER, TO_DATE(A.EVENT_DATE, 'DD-MON-YY'),B.SHORT_NAME FROM J_SUPPLY_CHAIN_HIST A, J_SOURCE B
[code]...
i created one view,i link two tables ,i can see the correct data in this view,but the result from front end application is not correct.i cant see the dates are correct.
1. I have 6 data bases, one process is running slowly, how do you find which data base that process belongs to?
2. I have 6 data bases, only one data base we have ASM instance, how do you find which data base has ASM?
I have made below query :
select job,case when deptno=10 then ename ELSE 'null' end e10
,case when deptno=20 then ename ELSE 'null' end e20
,case when deptno=30 then ename ELSE 'null' end e30
from emp
order by 1;
the output is:
JOB E10E20E30
ANALYSTnullSCOTTnull
ANALYSTnullFORDnull
CLERKnullSMITHnull
[code]...
but I want data in one row, i mean for example in case of JOB as 'CLERK' the output should be : i.e.
JOB E1 E2 E3
CLERK MILLER ADAMS JAMES
CLERK SMITH
the complete correct output which I want is:
JOB E1 E2 E3
ANALYST SCOTT
ANALYST FORD
CLERK MILLER ADAMS JAMES
[code]....
formulating sql query
Basically what I want is that I need to get desired result in such a way that, whenever Transaction type is Sales Order Issue, I want last TRANSACTION_COSTED_DATE of 'Intransit Shipment'
INVENTORY_ITEM_ID TRANSACTION_COSTED_DATE TRANSACTION_TYPE R
123 28-06-2012 21:36 Intransit Shipment
123 23-07-2012 01:25 Sales order issue 28-06-2012 21:36
123 30-07-2012 05:20 Sales order issue 28-06-2012 21:36
[Code]...
Lag with offset 1 doesn’t work as it will only go to previous row, What I want is that it should go to row above where transaction type is Intransit Shipment
Sample data and query I tried
with sampl_rownum_reset as
(select '123' inventory_item_id,
to_date ('28-Jun-2012 9:36:23 PM ', 'DD-MON-RRRR HH:MI:SS AM')
transaction_costed_date,
to_date ('28-Jun-2012 9:35:23 PM ', 'DD-MON-RRRR HH:MI:SS AM')
[Code]....
INSERT /*+ APPEND */
INTO sales_fact
SELECT *
FROM customer_sales_fact_staging
ORDER BY time_wh_id;
COMMIT;
[URL]
Suppose in the above case
SELECT *
FROM customer_sales_fact_staging
ORDER BY time_wh_id;
returns a huge number say more than 10 million or 50 million
will the commit work(one single transaction)
