Here are some examples of the most common options that Developers/DBAs use.

Export based on a query:
You can use thess options to export selected rows from a table. For example below I am exporting all records from orders table where id < 300    .

exp scott/tiger file=test1.dmp query=\”where id \< 300\” tables=orders statistics=none

Note above that I am using escape character “\”. This will be needed when you run your export in Unix/Linux environment.

Export selected tables:
You can use this option to export selected tables. The tables names are separated by comma (,). In the example below I am exporting orders and customers table data.

exp scott/tiger file=test1.dmp tables=orders, customers statistics=none

Exporting with Constraints
You can use this option to export data including constraints.

exp scott/tiger file=test1.dmp tables=orders, customers statistics=none constraints=y

Exporting with Feedback option:
You can use this option to see feedback while export is in progress.
Feedback (format feedback=n where n is a number). Note that each dot (.) represent the number you entered for feedback. So if you entered 100 then you get more dot.
exp scott/tiger file=test1.dmp tables=orders statistics=none constraints=y feedback=100
About to export specified tables via Conventional Path …
. . exporting table orders
……………
1569 rows exported

But if you enter 100 then you get just one .

exp scott/tiger file=test1.dmp tables=orders statistics=none constraints=y feedback=1000
About to export specified tables via Conventional Path …
. . exporting table orders
.
1569 rows exported
Export terminated successfully without warnings.

Exporting with indexes:
You can use this option to export data including indexes.

exp scott/tiger file=test1.dmp tables=orders statistics=none indexes=y

Exporting with Grants:
You can use this option to export data including grants..

exp scott/tiger file=test1.dmp tables=orders statistics=none grants=y

SQL*Loader is an Oracle supplied utility used to load data in Oracle tables easily with significantly
high performance. This utility is installed by default when you install Oracle Server software. It is
also installed when you install Administrative version of Oracle Client.

It is commonly used to load data from plain flat files (text files/plain data files) into database tables.
In it’s simplest form it requires a Control File, a Data File, the SQL*Loader executable and the Target
Table(s).

Control File: As the name suggests this file controls the loading of data in an Oracle table. Data type,
data length, data file separator (called field terminator), target table etc are specified in this file.

Data File: This is a plain text file separated by usually comma (“,”) or any other Oracle understandable
characters.

SQL*Loader: This is the sqlldr.exe file supplied by Oracle.

Target Tables(s): Data can be loaded into a single table or multiple tables.

Here is a demonstration of how SQL*Loader is used.
Suppose I have an Oracle table of structure

create table table1
(empno number,
fname varchar2(20),
lname varchar2(20),
age   number
);
I want to load data from a text file named mydatafile.txt which has the following data

1000,David,Selby,40
1001,Thomas,Train,31
1002,Nick,Junior,22
1003,Samantha,Dale,24
1004,Roman,Catfield,39
1005,Punnet,Gupta
1006,Dummy, Suname name does not exists, 10

To load the content of file mydatafile.txt into table table1 I’ll need to have a control file
which will look somewhat like this:

LOAD DATA
INFILE ‘mydatafile.txt’
INTO TABLE TABLE1
FIELDS TERMINATED BY ‘,’
TRAILING NULLCOLS
(
empno       INTEGER EXTERNAL NULLIF (empno=BLANKS)
, fname       CHAR NULLIF (fname=BLANKS)
, lname       CHAR NULLIF (lname=BLANKS)
, age         INTEGER EXTERNAL NULLIF (age=BLANKS)
)
Now I have got a Control File, a Data File, sqlldr.exe and a Target Table.

The next step is to actually load data into the table.

To do that you will login to the machine where sqlldr is installed. Depending on O/S you will either need a
Dos command prompt (MS Windows) or the Unix/Linux command prompt. I have run my test on Unix. I am currently
in the directory “/home/oracle/temp/” where I have got my control file and my data file. My Oracle user name
is scott (where my table table is was created) and password is tiger and Oracle_SID is ORADEV.

I execute the following command:

$sqlldr scott/tiger@ORADEV control=table1.ctl

And I get the following appear on my screen:

SQL*Loader: Release 11.1.0.7.0 – Production on Tue Jul 14 11:40:39 2009

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

Commit point reached – logical record count 7

“Record count 7″ means that it has processed 7 records, in does not necessarily mean it has inserted all
of the records in the table table1 as we wil lsee in a moment.
Looking at the directory I find there are two new files create called table1.log and mydatafile.bad. The
content of file table1.log looks like this:

[oracle@HTCD-DBOR01 chakma]$ cat table1.log

SQL*Loader: Release 11.1.0.7.0 – Production on Tue Jul 14 11:40:39 2009

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

Control File:   table1.ctl
Data File:      mydatafile.txt
Bad File:     mydatafile.bad
Discard File:  none specified

(Allow all discards)

Number to load: ALL
Number to skip: 0
Errors allowed: 50
Bind array:     64 rows, maximum of 256000 bytes
Continuation:    none specified
Path used:      Conventional

Table TABLE1, loaded from every logical record.
Insert option in effect for this table: INSERT
TRAILING NULLCOLS option in effect

Column Name                  Position   Len  Term Encl Datatype
—————————— ———- —– —- —- ———————
EMPNO                               FIRST     *   ,       CHARACTER
NULL if EMPNO = BLANKS
FNAME                                NEXT     *   ,       CHARACTER
NULL if FNAME = BLANKS
LNAME                                NEXT     *   ,       CHARACTER
NULL if LNAME = BLANKS
AGE                                  NEXT     *   ,       CHARACTER
NULL if AGE = BLANKS

Record 7: Rejected – Error on table TABLE1, column LNAME.
ORA-12899: value too large for column “QRS”.”TABLE1″.”LNAME” (actual: 28, maximum: 20)
Table TABLE1:
6 Rows successfully loaded.
1 Row not loaded due to data errors.
0 Rows not loaded because all WHEN clauses were failed.
0 Rows not loaded because all fields were null.
Space allocated for bind array:                  66048 bytes(64 rows)
Read   buffer bytes: 1048576

Total logical records skipped:          0
Total logical records read:             7
Total logical records rejected:         1
Total logical records discarded:        0

Run began on Tue Jul 14 11:40:39 2009
Run ended on Tue Jul 14 11:40:39 2009

Elapsed time was:     00:00:00.12
CPU time was:         00:00:00.03
Special importance should be given to the part

Table TABLE1:
6 Rows successfully loaded.
1 Row not loaded due to data errors.
0 Rows not loaded because all WHEN clauses were failed.
0 Rows not loaded because all fields were null.

It clearly says that “1 Row not loaded due to data errors”. So it means that we have not be
able to load all the records from the file. For this example it is quite clear which record
we have the problem with. But when there are thousands or even millions of records into
your flat file then it is not that easy to find out which records failed to load.

That’s why Oracle automatically created a filed called mydatafile.bad. In this case the file
will have the record “1006,Dummy, Suname name does not exists, 10″. As understandtable the
data in the lname field is too large to fit into the column called lname which has the size
of VARCHAR2(20). So it failed to load this record but loaded all other records in the file.

Have a look into the contents of the file table.log and you will find a line where it says

“ORA-12899: value too large for column “QRS”.”TABLE1″.”LNAME” (actual: 28, maximum: 20)”.

So this is a very clear error message and tells us why the record was rejected.

How this error will be hanlded is a decision to be taken care by your manager. Possible solution
may be increasing the column length (preffered in this case). Shortening the surname is another
option. But if you have failure on numerous records then you might have to remedy it in the
source and then re-create the data file again.