Questions About Chapter 2 in Oracle DB 10g: SQL Fundamentals II

Question

Hello,
first of all i'm glad to be a member of your forum. I have joined a beginner Oracle Course: Intro to SQL. I'm facing some problems understanding some concepts in Chapter 2 of Oracle Database 10g: SQL Fundamentals II text book. I got about 15 questions. However, i will only ask two questions at first. Since i'm a newbie, please answer it in a simplistic form. Excuse me if you see grammatical mistakes.
Dropping a column can take a while if the column has a large number of values. In this case it may be better to set it to be unused and drop it when the number of users on the system are fewer to avoid extended locks.
Questions:
"when the number of users on the system are fewer to avoid extended locks."
1. Can you explain this to me please?! fewer than before? fewer than? What if users kept increasing during the years! then this "fewer" may not happen until the company collapse!
2. Why do we need to use unused columns? When should we use unused columns?

Answer 1

Great! .... I got more questions, i just do not want to open a new same thread. Thus, i will just post the questions in here and i hope i will get help from experts...Please bare with me guys...The questions are numbered, unnumbered parts are information that helps you understand my question.
Note: just answer what you are willing to, one question or whatever you want. I'm not expecting to get all the answers from one member :)
Thanks for understanding
Page 2-7:
Certain columns can never be dropped such as columns that form part of the partitioning
key for a partitioned table or columns that form part of the primary key of an index- organized table.
Questions:
"columns that form part of the partitioning key for a partitioned table"
*1. Do they mean one table can be split into two different storage? What is the thing that*
link these both tables to make Oracle Server realize these two tables are actually one  table? Is is tablespace_name?
"columns that form part of the primary key of an index-organized table."
*2. Can you clarify the above sentence please*
*3. If i have set of columns that has large amount of data, i rather set them unused then*
drop them because the response time is going to be faster! I do not get it, can you
explain please? What i know is drop drops the column and release the disk space whilst
unused column make the columns useless and does not release disk space yet until we drop them, so
drop column does it in one step unlike taking the unused column process. In brief, i would like to know
why dropping unused columns that has large set of data is faster then dropping the column
directly...
Page 2-12
4. ALTER TABLE emp2 ADD CONSTRAINT emp_dt_fk
FOREIGN KEY (Department_id)
REFERENCES departments ON DELETE CASCADE);
The above query is written in text book. I think it should be written as
ALTER TABLE emp2 ADD CONSTRAINT emp_dt_fk
FOREIGN KEY (Department_id)
REFERENCES departments(dept_id) ON DELETE CASCADE;
Am i correct?
*5. Can you tell me what deferring constraints is in one sentence please? Why do we need it? When do we need it in real-life?*
*7. You can defer checking constraints for validity until the end of the transaction. A*
constraint is deferred if the system checks that it is satisfied only on commit. If a
deferred constraint is violated, then commit causes the transaction to roll back.
I do not understand the above paragraph, please explain. What i know is "end of
transaction" ends with ; or commit
Page 2-18
create table test1 (
pk NUMBER PRIMARY KEY,
fk NUMBER,
col1 NUMBER,
col2 NUMBER,
CONSTRAINT fk_constraint FOREIGN KEY (fk) REFERENCES test1,
CONSTRAINT ck1 CHECK (pk > 0 and col1 > 0),
CONSTRAINT ck2 CHECK (col2 > 0) );
-- "CONSTRAINT fk_constraint FOREIGN KEY (fk) REFERENCES test1"
*8. This is wrong isn't it? It references to test1 but no column specified.*
An error is returned for the following statements:
ALTER TABLE test1 DROP (pk); -- pk is a parent key.
*9. We can not drop it because we did not mention ON DELETE CASCADE. Am i right?*
ALTER TABLE test1 DROP (col1) -- col1 is referenced by multicolumn constraint ck1.
*10. I do not get it, can you explain please. col1 is not referenced, i see CHECK constraint is applied*
but no references made. Secondly, is ck1 considered multicolumn because it check two columns?
Or multicolumn here represents something else?
ALTER TABLE emp2
DROP COLUMN employee_id CASCADE CONSTRAINTS;
*11. This drop employee_id column and all its child. Correct?*
ALTER TABLE test1
DROP (pk, fk, col1) CASCADE CONSTRAINTS;
*12. This drops three columns and all its child if there are any. Correct?*
*13. Then what's the difference between ON DELETE CASCADE and CASCADE CONSTRAINTS?*
For example, What if employee_id in emp2 table definition does not have ON DELETE CASCADE,
will CASCADE CONSTRAINTS work? Please explain...
Page 2-22
When you are expecting a large data load and want to speed the operation. You may want
to disable the constraints while performing the load and then enable them, in which case
having a unique index on the primary key will still cause the data to be verified during
the load. So you can first create a nonunique index on the column designated as PRIMARY
KEY, and then create the PRIMARY KEY column and specify that it should use the existing
index.
Example:
1. create the table
create table new_emp
(employee_id number(6),
first_name varchar2(10)
2. create the index
create index emp_id_idx2 on new_emp(employee_id);
"You may want to disable the constraints while performing the load and then enable them"
so i suggest to load all data i want into new_emp.
3. create the primary key
alter table new_emp ADD primary key (employee_id) USING index emp_id_idx2;
What i understand is the following:
If we want to load large data into the new_emp, its better to create the table without any
constraints - in our case the constraint is primary key. After that, we create nonunique
index points to employee_id and then load data into new_emp. Finally, specify employee_id
as primary key using the nonunique index.
*14. Is my explanation correct?*
"in which case having a unique index on the primary key will still cause the data to be
verified during the load."
*15. Data to be verified against what? Is it to be verified whether its NULL or NOT NULL? I*
know primary key does not take NULL and every value must be unique.
After loading all data we want, what if i did
"alter table new_emp ADD primary key (employee_id);"
*16. Will i face any problems or inefficient process?*
I do not think we need step two, we could do the following:
1. create the table
create table new_emp
(employee_id number(6),
first_name varchar2(10)
"You may want to disable the constraints while performing the load and then enable them"
so i suggest to load all data i want itno new_emp.
2. create the primary key
alter table new_emp ADD primary key (employee_id);
*17. The above steps are as efficient as the three steps i mentioned above. The only difference*
is we let index be created implicitly. Right? If no, why?
Page 2-23
CREATE INDEX upper_dept_name_idx ON dept2(UPPER(department_name));
The following statement may use the index, but without the WHERE clause the
Oracle server may perform a full table scan:
select *
from employees
where UPPER(last_name) IS NOT NULL
ORDER BY UPPER (last_name);
"but without the WHERE clause the Oracle server may perform a full table scan"
*18. The above query let oracle server perform full table scan anyway! Right? It has to go*
through every field and check is it not null or not. I know we are using function-based
index but there are alot of not null last_name! so oracle server must scan one by one. If
we only had one not null field, then i would say Oracle server can point to that field
immediately by the aid of function-based index we created above. Can you clarify please...
Another related topic statement that i do not get it yet:
"The oracle server treats indexes with columns marked DESC as function-based indexes."
*19. The bove statements is so general. What if we have a column ordered by DESC order and we*
did not create any function-based indexes, will statement be true?!
Lets go back the above query:
ORDER BY UPPER (last_name);
*20. Its not DESC. To me, the above query does not flow with this statement "The oracle server treats*
*indexes with columns marked DESC as function-based indexes."?*
Page 2-27
Regarding FLASHBACK TABLE, you can invoke a flashback table operation on one or more
tables, even on tables in different schema. You specify the point in time to which you
want to revert by providing a valid timestamp. By default, database triggers are disabled
for all tables involved. You can override this default behavior by specifying the ENABLE
TRIGGERS clause.
"By default, database triggers are disabled for all tables involved. You can override this
default behavior by specifying the ENABLE TRIGGERS clause."
*21. What are database triggers?*
*22. About External Tables. What are external tables? When is it used in real-life? Why do*
we want External Tables?
Page 2-30
Oracle server provides two major access drivers for external tables. They are
ORACLE_LOADER access driver and ORACLE_DATAPUMP access driver. ORACLE_DATAPUMP used to
both import and export data using a platform-independent format.
"platform-independent format."
*23. What is the format? Is it .dat?*
Page 2-35
CREATE TABLE oldemp ( fname char(25), lname char(25) )
ORGANIZATION EXTERNAL
(TYPE ORACLE_LOADER
DEFAULT DIRECTORY emp_dir
ACCESS PARAMETERS
(RECORDS DELIMINATED BT NEWLINE
NOBADFILE
NOLOGFILE
FIELDS TERMINATED BY ',' (fname POSITION (1:20) CHAR, lname POSITION (22:41) CHAR)
LOCATION ('emp.dat') )
PARALLEL 5
REJECT LIMIT 200;
*24. Can you please explain the below part:*
ACCESS PARAMETERS
(RECORDS DELIMINATED BT NEWLINE
NOBADFILE
NOLOGFILE
FIELDS TERMINATED BY ',' (fname POSITION (1:20) CHAR, lname POSITION (22:41) CHAR)
*25. Can you please explain what is PARALLEL 5? and Why do we need it?*
Again, any help is appreciated...
Edited by: user11164565 on Jul 21, 2009 4:41 AM