DB2 Basics - Quick Refernce

Relational Database Management System(RDBMS)
Universal Database(UDB)

This is one of the sub systems in Mainframes.

Any number of Sub Systems can be created in Mainframes as per the requirements.

Hierarchy of DB2 Sub system:

TS – Table Space INS – Index Space

T – Table IND – Index

V – view

Maximum storage space for a Table Space is 64 million bytes.

SQL can be used to Create, Delete, and Update and Query the Objects

SQL queries can be executed by the following techniques

Application programming
Tools like QMF (Query Management Facility)

SPUFI (SQL Processor User File Input)

DATA TYPES :

Integer -- 4 bytes

Small int -- 2 bytes

Char(n) – N bytes

Varchar(n) – N+2 bytes

Graph(n) – 2n bytes

Vargraph(n)– 2N+2 bytes

Date – 10 bytes

Time – 8 bytes

Timestamp – 26 bytes

NORMALIZATION :

Arranging the data in the Database in organized manner.

1NF: Avoiding multiple values or set of values in one column.

2NF: Avoiding repeated rows by defining primary key.

3NF: Separating functionally dependent and non-functionally dependent columns

Primary key :-

Uniquely identified row
Which can be formed with single or multiple columns
Does not allow duplicate records
Cannot contain Null

Foreign key : -

Another identifier which can be used to build relationship between the tables
Must be the primary key of parent table with same data type & length
Can consists of single or multiple columns
Can contain Null or duplicate rows
Multiple foreign keys can be defined in one table
Foreign key should be defined at the time of defining child table in the create command by “WITH REFERENCES” option.

CREATE TABLE ITEM(

INO INTEGER,

INAME CHAR(15),

CNO INTEGER,

PRIMARY KEY IS INO,

FOREIGN KEY IS CNO

WITH REFERENCES CUST)

REFERENCE INTEGRITY:

The relationship between two tables which can be achieved by defining foreign key.

PRIMARY KEY	FOREIGN KEY
Cannot contain Null values or duplicate rows	Can contain Null values or duplicate rows
Cannot be updated	Can be updated
Can be defined as a foreign key in other table	which must be primary key of another table
only one primary key can be defined for one table	multiple foreign keys can be defined for one table

SQL(Structured Query Language)

DDL (Data Definition Language)

Create, alter, drop

DML (Data Manipulation Language)

Insert, update, select & delete

DCL (Data Control Language)

Grant, Revoke

TCL (Transaction Control Language)

Commit, Rollback

Some SQL Quries

Static SQL for Insert: Insert into cust(cno, cname, cloc) values (10, “xyz”, “hyd”)

Dynamic SQL for Insert: Insert into cust(cno, cname, cloc) values (v1, v2, v3)

v1,v2, v3 are host variables to be defined in working storage section.

Delete from cust

Delete from cust where cno = 20

Update cust set cname = “ xyz” where cno = 20

Select cno,cname from cust

Select * from cust

Select * from, cust where cno = v1

Select * from cust wehre cno=v1 and cname =v2

Select * from cust where cno between 20 and 60

Select * from cust where cname like “%y%”

Column functions:

Select max(sal) from emp

Select min(sal) from emp

Select avg(sal) from emp

Select sum(sal) from emp

Above statement returns Null values if no row exits for specified condition

To avoid duplicate rows : select distinct cno,cname from cust

To get total no. of rows : select count(*) from cust

Above statement returns Zeros if no row exits for specified condition

SUBQUERY:

Query within Query
First inner query executes & out query executes based on the result of inner query
Max of 15 sub queries can be coded
To simplify sub queries, logic can be built with combination of COBOL + SQL statements

To retrieve second maximum salary from emp table:

Select max(sal) from emp where sal <(select max(sal) from emp)

To retrieve third maximum salary from emp table:

Select max(sal) from emp

where sal < (select max(sal) from emp

Where sal < (select max(sal) from emp))

CO-RELATED SUBQUERY:

For every row of outer query, inner query must executes at least once
First outer query executes & then inner query executes
Practical examples : to fine top 2,3 or n salaries

Select a. sal from emp a

where 0 = (select count(*) from emp b

Where a.sal < b.sal)

-- max
– 2^nd max
– 3^rd max
– 4^th max

DCLGEN :

Declaration Generator . a tool to generates the equivalent COBOL variables.

Which can be used to generate host variables with equivalent data types of DB2 columns.

DB2 Table DCLGEN COBOL

Host variables:

Can be used to pass the data from cobol program to DB2 table or DB2 table to COBOL program.
When host variables are coded with sql statements it must be prefixed with : like :hv-cname.
Table name must be supplied as input to DCLGEN & partition dataset should be as output.
After creating DCLGEN variables which must be copied to application program in WORKING-STORAGE SECTION by using include command i.e.

Exec sql

Inlcude custDCL

End-exec.

Include & copy have the same functionality

SQLCODE :

Predefined numeric number which can be used to check SQL statements for successful , unsuccessful execution.
SQLCODE can be stored in SQLCA(SQL Communication Area)
Copy SQLCA in WORKING-STORAGE SECTION
System defined variable
Evaluate or if statement must be coded to check the SQLCODE immediately after SQL statement.
SQLCODE =00 ---- successful

= +100 --- end of table or record not found.

Sample program:

WORKING-STORAGE SECTION.

EXEC SQL

INCLUDE SQLCA

END-EXEC

EXEC SQL

INCLUDE CUSTDCL

END-EXEC.

01 WS-SQL-CODE PIC S9(4)

88 88-SUCCESS VALUE 00

88 88-NOTFOUND VALUE 100

88 88-FORIENG KEY VOILATION VALUE –532

88 88- MULITPLE ROW VALUE –811

PROCEDURE DIVISION.

UPDATE CUST

SET CNAME = :HV-CNAME

WHERE CNO=:HV-CNO

MOVE SQLCODE TO WS-SQLCODE.

EVALUE WS-SQL-CODE

WHEN 88-SUCCESS

DISPLAY “SUCCESSFULLY UPDATED”

WHEN 88-NOTFOUND

DISPLAY “ RECORD NOT FOUND”

WHEN 88-FOREIGNKEYVOILATION

DISPLAY “ FOREIGN KEY VOILATION”

WHEN OTHER

DISPLAY “ ERROR OCCURRED IN UPDATE”

STOP RUN

END-EVALUATE.

STOP RUN.

CURSOR:

To retrieve multiple rows for a given condition.

Let us take the following example:

Exec sql

Select cno,cname,cloc

into :hv-cno,:hv-cname,:hv-cloc

from cust where cloc =:hv-cloc

end-exec.

If the condition satisfy for one row it executes successfully. If the condition satisfy for multiple rows it wont work. It returns –811 as SALCODE. For this we use cursors.

Cursors can be used to retrieve multiple rows for a given condition
Cursor cycle is Declare ---> Open ----> Fetch -----> Close
Declare: declares or define name for cursor against a table

Can be coded in working-storage section or procedure division

For better readability code in working-storage section.

Open: can be used to open a cursor with rows for a given conditions inbuffer.

Retireves data in to buffer

Must be coded in the procedure division only

Where condition value must be supplied before opening a cursor.

Fetch: can be used to retrieve rows one by one from buffer into application prog.

Which must be coded in procedure divison after open.

Must be coded with hostvariables

No of host variables in fetch & no of columns in the declare must be same

Canbe executed multiple times using perform. i.e. till EOT or record not found which can be identified by SQLCODE = 100

Close : used to close the cursor

Must be coded in procedure division only

Must be executed after open statement.

Practical examples : Can be used to retrieve the data based on loc, date, products.

EXEC SQL

DECLARE C1 CURSOR FOR

SELECT CNO,CNAME FROM CUST

WHERE CNAME=:HV-CNAME

END-EXEC.

EXEC SQL

OPEN C1.

END-EXEC.

PERFORM UNTIL SQLCODE= 100

EXEC SQL

FETCH C1 INTO :HV-CNO,:HV-CNAME

END-EXEC

END-PERFORM.

EXEC SQL

CLOSE C1

END-EXEC

For Update of where current of:

Which can be used to update row by row when multiple rows are satisfied.
Before update cursor has to be declared with for update of column option.
Where current of cursor name option must be used with update command
Withhold: this option can be used to remain cursors open even after commit statement.
Must be coded with cursor statement

EXEC SQL

DECLARE C2 CURSOR WITH HOLD FOR

SELECT CNO,CNAME FROM CUST

WHERE CNAME=:HV-CNAME

FOR UPDATE OF CNAME

END-EXEC.

EXEC SQL

OPEN C1.

END-EXEC.

EXEC SQL

FETCH C2 INTO :HV-CNO,:HV-CNAME

END-EXEC

EXEC SQL

UPDATE CUST SET CNAME=”ABC” WHERE CURRENT OF C2.

EMD=EXEC.

EXEC SQL

CLOSE C1

END-EXEC

INDEX:

Index allows duplicate values
unique index doesn’t allow duplicate rows
cross reference between index table & table is called clustered index.
Create index in1 on cust(cno)

PRIMARY KEY	INDEX	UNIQUE INDEX
Uniquely identified row	Record identified based on the index	Records identified based on the index
No duplicated rows, no null values	Duplicate rows, null values are allowed	No duplicate rows
Can consist of single or multiple columns	Dan consist of single or multiple columns	Can consist of single or columns
This will be stored in SYSKEYS.	This is stored in SYSINDEX	This is stored in SYSINDEX

VIEWS:

CREATE VIEW CVIEW(VCNO,VCNAME,VCLOC) AS

(SELECT CNO,CNAME,CLOC FROM CUST WHERE

CNAME LIKE “%X%)

Logical representation of the table
Stored in virtual memory
Can be derived from single table or multiple tables
Views are updateable if they are derived from single table without any column functions , group by
Multiple views can be generated from single table.
Views are stored in sysviews

Advantages of Views:

Data security
Data correctness
Logical data independence
Part of the information can be visible to the sers
Accessing can be faster.

DELETE RULES:

Delete rules can be applied for delete command against Database.
Delete rules are 3 types

UNION:

UNION is used to concatenate rows into one table from single or multiple tables.
Rules : no. of columns & data type of both the queries must be same. column may be different
UNION can be used to eliminate duplicate rows
UNION ALL retrieved duplicate rows also.

SELECT CNO,CNAME FROM CUST WHERE CNO=10

UNION/UNIONALL

SELECT CNO,CNAME FROM ITEM WHERE INO=20

JOINS:

JOINS can be used to concatenate columns from one table or multiple tables.
JOIN types are :

EXPLAIN :

It can be used to evaluate the performance of SQL queries.

It can be used to tune SQL queries.

Input is SQL queries and output is plan-table.

For every SQL query one plan-table will generate.

All plan-tables are stored in physical seq file.

Plan table

Query

block no

Table

Name

No. of

Cols

index

no.of

indexs

owner

join type

groupby

Cpu

time

Cust

In1

Custc

Self

10 min

DB2 CATALOG:

Consists of Table pace, Index space, Index, unique index, Views, Alias, synonyms, keys.
When we create table, the details of table are entered in Systable automatically.

SysIBM.SYSTABLE

Table Name	No.of cols	Owner name	Created by	Created date	Created time
Cust	10	Abc	Xyz	02-apr-2004	0850
Item	15	Mno	Rst	06-apr-2004	1020

SysIBM.SYSINDEX

Index name	Table Name	No.of cols	Owner name	Created by	Created date	Created time
In1	Cust	10	Abc	Xyz	02-apr-2004	0850
In2	Item	15	Mno	Rst	06-apr-2004	1020

SysIBM.SYSCOLS

Col name	Table name	Index name	Primary key	Foreign key
Cno	Cust	In1	Cno	-----
Cname	Cust	In1	Cno
Cloc	Cust	In2	cno
Ino	Item		Ino	Cno
Iname	Item		Ino	Cno
Ides	Item		Ino	cno

SysIBM.SYSKEYS: All primary & foreign keys.

Grant table syscols to all

Grant table syscols(select/delete/update) to user1,user2

Revoke table syscols from all.

CATALOG:

SYSTABLE, SYSCOL, SYSKEYS, SYSINDEX, SYSPKS, SYSFKS, SYSALIAS, SYSSYNONYMS, SYSINDEX, SYSVIEWS,SYSTABLESPACE, SYSINDEXSPACE.

PRECOMPILATION PROCESS:

Pre compiler takes COBOL+DB2 program as input & generates DBRM which will be stored in userdefined PDS as separate member of Recln=80

DSNHPC --- IBM supplied utility used for precompilation.

Precompiler functions:

Separates SQL & COBOL statements
Check SQL syntaxs
Replace all SQL statements with host language call statements in the compiled program.
Which generates timestamp tokens

BIND:

BIND takes DBRM as input & generate package & application plan. The package will be loaded to the directory. Plan will be loaded to sysplans.

Bind functions:

Checks authorization for SQL statement
Checks the syntax errors of SQL statements like

Missing co name in the select list & used in order by & group by
Mismatch columns host variables
Data type mismatch of columns & host variables
Indicator variables not declared
Data truncation.

BIND SUBCOMPONANTS/PARAMETERS:

OPTIMIZER:
- It generates optimized access path by analyzing the statistics of SQL statements which will be stored.
- RUNSTATS utility is one of the ISPF panel option which is stored in DB2 defaults option.
- Optimized path is stored in package which is not executable module.

ISOLATION LEVEL:

Which can be used to lock the database at the time of executing SQL statements.

Cusrsor stability(CS): It acquires the address of a row. Sets the pointer to a specified row based on SQL query & acquires the lock against that row. Then releases the klock after the transaction before commit.

Repeatable Read(RR): which acquires the address of a row & acquire lock against the page(1 page -4024 bytes) & then released the lock after the commit statements.

Default is RR.

RUNTIME SUPERVISOR:

Which is to oversee execution of SQL statements.

Statistics like no of tables, columns, indexes, keys

PLAN/APPLICATION PLAN:

It consists of executable module which is actual output of SQL statements which must be specified in the RUNJCL to execute SQL queries if the program is batch program. If the program is online which must be specified in RCT. Application plan will be loaded to load module with time stamp tokens.

COBOL COMPILATION:

The compiler takes COBOL statement as input to generate object program, & loaded to the load module by line/edit with time stamp tokens.

UTILITIES USED:

DSNHPC : system utility pre compiler.

IKJEFT01 or IKJEFT01B --- BIND /Run Cob-DB2 Program

IGYCRCTL or IKFCBLOO --- COBOL compilation

IEWL or HEWL --- link/edit

INTERVIEW QUESTIONS: (watch this space for more info on below Qns)

What is RI ? where did u use RI in your project? Explain with example?
What is the difference between primary key, foreign key, index & unique index?
Can we create index when table has duplicate rows?
Can we create unique index when table has duplicate rows?
Can we create index or unique index on empty table?
What happens to the package when index is dropped? What’s the process or steps to be taken?
How to delete package?
Where package is stored? How to retrieve package?
Difference between plan & package?
What are the steps to be taken when SQL statements are changed without changing any COBOL statements?
Do we need to pre compile when index is dropped?
Can we bind when table is dropped?
Can optimized access path consist of multiple DBRMS.
What is the significance of timestamp tokens?
What is the significance of normalization?
Where do we specify run program, plan name, library & DB2 subsystem
IBM supplied utility to run COBOL + DB2
What is difference between DB2 & COBOL files & give some example for COBOL files & DB2 tables related to your project?
Can we load data from sequential file to table or table to sequential file?
What are the steps to be followed to develop DB2+ COBOL program?
Can we prepare a program/compile when DB2 system is down?
How to identify DB2 test or production system by seeing run JCL?
What is the output of explain?
What is the difference between correlated sub query & sub query?
How to find 4^th max sal?
How to find nth max sal?
How to evaluate SQLcodes & where it is stored?
How to count total no of unique rows from the table?
How to sum repeated rows?
How to write a cobol program for above query? Retrieve row by row & use cobol logic to sum repeated rows?
What is the significance of DCLGEN?
Where DCLGEN is stored?
Difference between join & union?
difference between UNION & UNIONALL?
Can be have different col names in union?
How do u evaluate/tune/analyze SQL queries?
What are the JCL utilities for compile, pre compile, bind & link edit?
Wha5t is the significance of isolated levels?
Can we alter foreign key?
Can we alter primary key?
Can we alter data type & length?
What are the equivalent cobol variables for varchar?
What is the time stamp & its format?

Mainframe For Dummies

Search This Blog

DB2 Basics - Quick Refernce

Comments

Popular posts from this blog

How to Solve SOC7 Abend - with screen shots

NULL VALUES and NULL INDICATORS in DB2

Mainframe Written test @ Accenture