$25
CSCI235 -Ass 1 - Solved
Normalization of relational tables
Consider the following conceptual schema of a sample database domain where the students borrow the books from a library. A book is described by a call-number that uniquely identifies each copy of a book kept in a library. A library purchases several identical textbooks for the subjects enrolled by the larger number of students. A book is identified by ISBN and it is described by a title and publisher.
The students are described by a student number, first name and last name. A date is recorded when a student borrows a book.
A database designer made few mistakes at both conceptual modelling stage and logical design stage (i.e. transformation of a conceptual schema into the relational schemas).
At present, the relational schemas created by a database designer are the following.
BOOK(call-number, ISBN, title, publisher) primary key = (call-number)
STUDENT(student-number, first-name,last-name) primary key = (student-number)
BORROW(call-number, ISBN, student-number, borrow-date) primary key = (call-number, student-number)
Your task is to use the analysis of functional dependencies and normalization of relational schemas to find the highest normal form valid for each one of the relational schemas listed above. If a relational schema is not in BCNF then you must decompose it into the relational schemas in BCNF.
The process of normalization of relational schemas must be performed in the following way. First, find functional dependencies valid in a relational schema. Next, find the minimal keys. Next find, the highest normal form valid for a relational schema, and finally if a normal form found is not BCNF decompose a schema into BCNF.
Repeat such process for every relational schema listed above.
Please remember, that both conceptual schema and relational schemas are not completely correct, so be very careful when finding functional dependencies. Using a conceptual schema and relational schemas only is like "walking over a mine field".
!
Task 2 (3 marks) Normalization of relational tables
(1) Consider the following relational schema in BCNF.
ROOM(bldg-number, room-number, room-area) primary key = (bldg-number, room-number)
Add to a relational schema ROOM a meaningful attribute such that after addition the relational schema will be at most in 1NF and not in 2NF. Prove, that after addition of an attribute the relational schema is not in 2NF.
(2) Consider the following relational schema in BCNF.
FLIGHT(flight-number, from-city, to-city) primary key = (flight-number)
Add to a relational schema FLIGHT a meaningful attribute such that after addition the relational schema will be at most in 2NF and not in 3NF. Prove, that after addition of an attribute the relational schema is not in 3NF.
(3) Consider the following relational schema in BCNF.
ORDER(order-number, supplier) primary key = (order-number)
Add to a relational schema RDER a meaningful attribute such that after addition the relational schema will be at most in 3NF and not in BCNF. Prove, that after addition of an attribute the relational schema is not in BCNF.
(2), and (3). For each one of the cases listed above provide a proof that after addition of a meaningful attribute a relational schema is in a required normal form.
Task 3 (5 marks) Indexing
Prologue
Download the files dbcreate.sql, dbload.sql, and dbdrop.sql included in a section SAMPLE DATABASES on Moodle. To drop a sample database, process a script dbdrop.sql. To create a sample database, process as script dbcreate.sql. To drop a sample database, process a script dbdrop.sql. To load data into a sample database, process as script dbload.sql.
Connect to Oracle database server and process the following SQL statement that saves a query processing plan for a given SELECT statement in PLAN_TABLE.
EXPLAIN PLAN FOR
SELECT R_COMMENT
FROM REGION
WHERE R_NAME = 'Africa';
Next, process the following SELECT statement to display a query processing plan stored in PLAN_TABLE.
SELECT * FROM TABLE(DBMS_XPLAN.DISPLAY);
Among the others, you should get the following results.
PLAN_TABLE_OUTPUT
----------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
----------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 105 | 3 (0)| 00:00:01 |
|* 1 | TABLE ACCESS FULL| REGION | 1 | 105 | 3 (0)| 00:00:01 |
----------------------------------------------------------------------------
Predicate Information (identified by operation id):
1 - filter("R_NAME"='Africa')
A line TABLE ACCESS FULL| REGION in a plan given above indicates that a database system plans to read entire table REGION to compute the query.
Next, create an index on a column R_NAME in a relational table REGION.
CREATE INDEX REGION_IDX ON REGION(R_NAME);
Again, process the following SQL statements that save a query processing plan for the same SELECT statement as before in PLAN_TABLE and display a query processing plan stored in PLAN_TABLE.
EXPLAIN PLAN FOR
SELECT R_COMMENT
FROM REGION
WHERE R_NAME = 'Africa';
Among the others, you should get the following results.
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 105 | 1 (0)| 00:00:01 |
| 1 | TABLE ACCESS BY INDEX ROWID BATCHED| REGION | 1 | 105 | 1 (0)| 00:00:01 |
|* 2 | INDEX RANGE SCAN | REGION_IDX | 1 | | 1 (0)| 00:00:01 |
--------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
2 - access("R_NAME"='Africa')
INDEX RANGE SCAN | REGION_IDX
This time a database system plans to use an index REGION_IDX created a moment ago to process the same query. Note, a line in a plan given above means, that a database system plans to vertically traverse an index REGION_IDX to find the identifiers of rows that satisfy a condition R_NAME =
'Africa';. A line TABLE ACCESS BY INDEX ROWID BATCHED|
REGION means, that the system plans to group row identifiers pointing to the same data blocks in order to minimize the total number of read block operations and to access data blocks of a relational table REGION pointed by the row identifiers to find the values of an attribute R_COMMENT.
Conclusions
EXPLAIN PLAN statement of SQL can be used to get information about a processing plan created by a query processor for a given SELECT statement. A query processing plan provides information on whether and index created earlier will be used for processing of SQL statement. We shall use EXPLAIN PLAN statement to check whether an index created to speed up SELECT statement will be used for processing of the statement.
To drop an index, process a statement
DROP INDEX REGION_IDX;
No report is expected from processing of SQL statements given above.
Problem
Your task is to find what indexes should be created to speed up processing of SELECT statements listed below. You are expected to create one index for one SELECT statement. To simplify the problem, assume that any index, later on used by a query processor to speed up processing of SELECT statement, will do.
(1)
SELECT O_ORDERDATE, O_CUSTKEY
FROM ORDERS
WHERE O_ORDERDATE > '31-DEC-1994' AND
O_ORDERDATE < '01-JAN-1996';
(3)
SELECT L_PARTKEY, L_SUPPKEY
FROM LINEITEM
WHERE ( L_TAX IN (0.02, 0.06) OR L_DISCOUNT> 0.4 ) AND
L_EXTENDEDPRICE = 7232;
(3)
SELECT DISTINCT PS_AVAILQTY, PS_SUPPLYCOST, PS_SUPPKEY
FROM PARTSUPP;
(4)
SELECT P_BRAND, AVG(P_RETAILPRICE)
FROM PART
GROUP BY P_BRAND;
(5)
SELECT *
FROM SUPPLIER
ORDER BY S_NAME, S_NATIONKEY;
Implement SQL script solution3.sql such that for each one of SELECT statements given above the script performs the following actions.
(i) Find and list a query processing plan for SELECT statement without an index.
(ii) Create an index.
(iii) Find and list a query processing plan for SELECT statement with an index. (iv) Drop an index.
When ready process SQL script file solution3.sql and save a report from processing in a file solution3.lst.
Your report must include a listing of all PL/SQL statements processed. To achieve that put the following SQLcl commands:
SPOOL solution3
SET ECHO ON
SET FEEDBACK ON
SET LINESIZE 300
SET PAGESIZE 200
at the beginning of SQL script and
SPOOL OFF
at the end of SQL sc
Task 4 (6 marks)
Implement an anonymous PL/SQL block that lists in a format listed below the names of regions (attribute R_NAME in a relational table REGION), and the names of at most 3 nations (attribute n_nation in a relational table NATION) included in each region. The names of regions must be sorted in ascending order and the names of nations must be sorted in descending order. All name must be listed in uppercase letters.
REGION-NAME-1 : NATION-NAME-1 NATION-NAME-2 NATION-NAME-3
REGION-NAME-2 : NATION-NAME-1 NATION-NAME-2 NATION-NAME-3
REGION-NAME-3 : NATION-NAME-1 NATION-NAME-2 NATION-NAME-3
... ... ... ... ...
To list information retrieved from a sample database use PL/SQL package DBMS_OUTPUT. It is explained in the Cookbook, Recipe 7.1 How to start programming in PL/SQL how to use DBMS_OUTPUT package. Remember about SET SERVEROUTPUT ON at the beginning of a script file that contains your anonymous PL/SQL block.
Your implementation must use at least one cursor and at least one exception handler. In fact, such constraints make your implementation easier.
To test your solution put an implemented anonymous PL/SQL block into SQL script file solution4.sql and process the script.
Your report must include a listing of all PL/SQL statements processed. To achieve that put the following SQLcl commands:
SPOOL solution4
SET SERVEROUTPUT ON
SET ECHO ON
SET FEEDBACK ON
SET LINESIZE 200
SET PAGESIZE 400
SET SERVEROUTPUT ON
at the beginning of SQL script and
SPOOL OFF
at the end of SQL script.
