DB2 UDB FOR Z/OS

Overview
Embedded SQL in DB2

An Introduction

DB2-DM-1
 Host Languages

 COBOL
 PL/1
 C
 Assembler
 FORTRAN
 Embedded SQL

 Delimit all SQL statements

 Declare a Communication Area

 Describe Host Variables

 Code SQL statements to access DB2 data

 Handle Exceptional Conditions

DB2-AP-2
 DELIMITING SQL Statements

DELIMITING SQL STATEMENTS

 Coding SQL statements in a COBOL application requires beginning and ending delimiters

 The delimiters are EXEC SQL & END – EXEC

 Whether or not a period is required after the END – EXEC is based on the logic of the
COBOL coding.

FORMAT

EXEC SQL
an SQL statement
END – EXEC.
 SQL COMMUNICATION ARE (SQLCA)

SQL COMMUNICATION ARE (SQLCA)

 DB2 program communication is accomplished through the SQL Communication

Area: SQLCA
 SQLCA provides the fields set by DB2 after the SQL statement execution;

The return code placed in the field SQLCODE indicates the success, failure or
execution encountered
 DECLARING SQLCA

The COBOL description of SQLCA:

01 SQLCA.
05 SQLCAID PIC X(8) VALUE ‘SQLCA’.
05 SQLCABC PIC S9(9) COMP.
05 SQLCODE PIC S9(9) COMP.
05 SQLERRM.
49 SQLERRML PICS9(4) COMP.
49 SQLERRMC PIC X(70).
05 SQLERRP PIC X(8).
05 SQLERRD OCCURS 6 TIMES
PIC S9(9) COMP.
05 SQLWARN.
10 SQLWARN0 PIC X(1).
10 SQLWARN1 PIC X(1).
10 SQLWARN2 PIC X(1).
10 SQLWARN3 PIC X(1).
10 SQLWARN4 PIC X(1).
10 SQLWARN5 PIC X(1).
10 SQLWARN6 PIC X(1).
10 SQLWARN7 PIC X(1).
05 SQLEXT PIC X(8).
 DECLARING SQLCA

 An SQLCA must be declared in an WORKING – STORAGE SECTION of every

DB2/SQL COBL program

 Declaration of SQLCA can be done in two ways:

 Coding the COBOL description of SQLCA directly

 Coding the SQL statement:

EXEC SQL
INCLUDE SQLCA
END-EXEC.
 SQLCA FIELDS AND THEIR PURPOSE

SQLCAID Always set to a value “SQLCA”;

Useful when debugging using a dump
SQLCABC Specifies the length of the SQLCA area;
A full word integer with a value of 136
SQLCODE A full word integer field containing the return code passed by DB2
relating to
the execution of the latest SQL statement;

SQLCODE Characteristics of different return codes:

0 Statement executed successfully

<0 An error occurred during statement execution;

>0 The SQL statement executed successfully, but an exceptional

condition occurred;

+100 No data exists to process or end of results table reached

 SQLCA FIELDS AND THEIR PURPOSE

SQLERRM Gives a brief message about any error condition

encountered;
Provides the names of object and other information

SQLERRD(n) n takes a usual value of 6;

SQLERRD (3) is the only variable in this array that is
pertinent to a
program’s execution;
The value in SQLERRD(3) is set to reflect the number of
rows
INSERTED, or DELETED
 SQLCA WARNING FLAGS

One (or more) of these is set when an SQL statement is executed, and something out of
the ordinary is detected by DB2.

SQLWARN0 Set to a ‘W’ when one of the other flags is also set to a ‘W’

SQLWARN1 Set to a ‘W’ when column values are truncated being placed
into a host variable

SQLWARN2 Set to a ‘W’ when a NULL value was eliminated from any
built-in-functions.

SQLWARN3 Set to a ‘W’ when there is an unequal number of host variables

between the SQL statement and the TABLE/VIEW being
processed

SQLWARN4 Set to a ‘W’ when an UPDATE or DELETE does not contain a

WHERE clause

SQLWARN5 Set to a ‘W’ if the program issued an SQL statement that

applies
only to SQL/DS

SQLWARN6 & Not being used

SQLWARN7
 HOST VARIABLES

 Are defined in the WORKING-STORAGE section or the LINKAGE SECTION

of the DATA DIVISION just like any other COBOL data items.
01 HS – EMPJOB.
02 HV – EMPNO PIC X(6).
02 HV – JOBCODE PIC S9(3).
01 HV – EMPID PIC X(6).

 Are preceded by a COLON when used in SQL statements to signify that

they are host variables and not SQL variables.
EXEC SQL
SELECT EMPNO, JOBCODE
INTO ; HV – EMPNO, : HV – JOBCODE
FROM EMPLOYEE
WHERE EMPNO = : HV – EMPID
END – EXEC.

 Are just regular COBOL data items when used in a non – SQL COBOL
statement

MOVE INPUT – EMPID TO HV – EMPID.

 SQL and COBOL Data types

 SMALLINT PIC S9(04) COMP

 INTEGER PIC S9(09) COMP

 DECIMAL(p,s) PIC S9(p-s)V9(s) COMP-3

 CHAR(n) PIC X(n)

 VARCHAR(n) 01 var .
49 var1 PIC S9(04) COMP.
49 var2 PIC X(n) .
 HOST VARIABLES

 Are just the variables of the host programming language

 Are variables that provide the following functions:

 Selection Criteria as part of the WHERE clause predicates

 Data Receiving Area as variable with the INTO clause

 Data Variables in SELECT, INSERT or UPDATE statement.

 Columns and the corresponding host variables data types must be

compatible
 HOST STRUCTURE

 Is a group level of host – variables defined in the DATA DIVISION of

the program.

 Matches the view or table referenced in the application

 Relieves the programmer from naming each variable in the SQL

statement

 Can only be two levels, except in case a column is defined as variable

length;

 In that case, the third level must be a level 49

 Structure can itself be part of a structure

 INDICATOR VARIABLES

 A variable of SMALLINT data type associated with a host variable

 Must be declared.
 Is preceded by a colon and directly follows its associated host variables, when
used in an SQL statement.

EXEC SQL
SELECT HIREDATE’
INTO :HV – HIREDATE : IV – HIREDATE
FROM EMPLOYEE
WHERE EMPNO = :HV – EMPNO
END – EXEC.

Note * IV – HIREDATE is the indicator variable associated with the host

variable
HV - HIREDATE
 INDICATOR VARIABLES

 DB2 returns one of the following values to an indicator variable:

if the vale returned to the associated host variable by DB2

0 is NOT NULL, not truncated

-1 is NULL
>0 is a truncated character string; the value in the indicator variable is the
length of the character string before truncation.
 INDICATOR VARIABLES

 An Indicator Variables can be used to:

 Indicate if its associated host variable has been assigned a NULL value
 Indicate if a returned character string has been truncated
 Set NULL value in the column
MOVE – 1 TO IND – HIREDATE.

EXEC SQL UPDATE EMPLOYEE

SET HIREDATE =:HV – HIREDATE:IND – HIREDATE
WHERE EMPNO =:HV – EMPNO
END – EXEC.
 INDICATOR STRUCTURE

* Is defined to support a Host Structure

01 H-STRUCTURE.
02 HV-EMPNO PIC..
02 HV-HIREDATE PIC..
02 HV-SALARY PIC..
01 I-STRUCTURE.
02 IV PIC S9(4) COMP
OCCURS 3 TIMES.
EXEC SQL
SELECT EMPNO, HIREDATE, SALARY
INTO :H-STRUCTURE:I-STRUCURE
FROM EMPLOYEE
WHERE EMPNO = :HV-EMPNO
END-EXEC.
 DECLARING DB2 TABLES

• DB2 tables to be used in a COBOL program must be declared in the WORKING-

STORAGE SECTION

* Two ways of declaring tables:

 Coding the table declaration directly, using the DECLARE TABLE statement
EXEC SQL
DECLARE TDEPT TABLE
(DEPT NO CHAR(3) NOTNULL,
DEPTNAME VARCHAR(36) NOTNULL,
MGRNO CHAR(6) NOTNULL,
ADMRDEPT CHAR(3) NOTNULL)
END-EXEC.

 Using DCLGEN (Declare Generation) to generate the table declaration and table
copying the same into the program in the WORKING-STORAGE SECTION using the
INCLUDE statement

EXEC SQL
INCLUDE DCLDEPT
END-EXEC.

DCLTDEPT is the member of a partitioned data set generated by DCLGEN and

 DECLARING DB2 TABLES

DCLGEN

• The DCLGEN panel is used to generate the table or view program

declarations
• DCLGEN can be accessed from ISPF through DB21, directly from TSO
with the DCLGEN command or in a batch mode with JCL
• The 1st part of the output is the TABLE declaration. Declaration the
table in the program is optional, but it gives the following advantages:
• DOCUMENTATION – the structure of the TABLE and the data type of the
columns are specified
• The DB2 precompiler validates that the correct column names and data
types are used in the SQL statements. Otherwise, this validation will
not occur until BIND time.
• The 2nd part of the output is the host language data structure.
DCLGEN generates this structure with appropriate data types and
therefore reduces the chances of error while coding
 CODING SQL TO MANIPULATE DB2 DATA

Considering the following SELECT statement:

EXEC SQL
SELECT EMPNO, SALARY
INTO :HV-EMPNO, :HV-SALARY
FROM EMPLOYEE
WHERE EMPNO =:HV-EMPID
END-EXEC.

* Each evaluation of the statement returns at most ONE row from the
table EMPLOYEE

* DB2 places the retrieved data directly into the host variables

* The retrieved data is now ready for the program to use

Note One can include the INTO clause in a select statement and set
up host variables to receive data directly from DB2 only when
the SELECT returns at most one row at a time.
 HANDLING EXCEPTIONAL CONDITIONS

Exceptional conditions, returns after the execution of an SQL statement, can

be
handled by the SQL WHENEVER declarative statements in general

EXEC SQL
WHENEVER {SQLWARNING {GOTO LABEL
/SQLERROR /GOTO LABEL
/NOT FOUND} /CONTINUE}
END – EXEC.

* SQLWARNING Condition exists when

SQLWARN0=‘W’
* SQLERROR Condition exists when
SQLCODE<0
* NOTFOUND Condition exists when
SQLCODE=100
* GO TO LABEL Passes control to COBOL routine label
* CONTINUE Passes control to next program instruction

NOTE The WHENEVER statement MUST precede the SQL statements it

is to affect
 ALTERNATE ERROR HANDLING
ECHNIQUES

* Omitting a WHENEVER statement is the same as coding it with

CONTINUE;
the next program instruction gets the control

* One may explicitly test the data fields in SQLCA

EXAMPLE

Explicitly testing the value of SQLCODE:

IF SQLCODE = 100
PERFORM OUTPUT – ROUTINE.

Explicitly testing the value of SQLWARN1:

IF SQLWARN1 = ‘W’
DISPLAY ‘VALUE US TRUNCATED’.
 A SIMPLE DB2/SQL COBOL PROGRAM

IDENTIFICATION DIVISION.
PROGRAM-ID. SAMPLE1.
*
ENVIRONMENT DIVISION.
*
DATA DIVISION.
WORKING-STORAGE SECTION.
01 WS-VARS.
05 CNAME PIC X(20).
05 FAT-NAME PIC X(20).
05 M-NAME PIC X(20).
05 DOB PIC 9(6).
05 STATE PIC X(10).
*
EXEC SQL
INCLUDE SQLCA
END-EXEC.
*
EXEC SQL
INCLUDE ORPHAN
END-EXEC.
 A SIMPLE DB2/SQL COBOL PROGRAM
PROCEDURE DIVISION.
0001-MAIN-PARA.
DISPLAY “Enter the Name of the Child”.
ACCEPT CNAME.
DISPLAY “Enter the Child’s Father’s Name; Enter Blank if Unknown”.
ACCEPT FAT-NAME.
DISPLAY “Enter the Child’s Mother’s Name”.
ACCEPT DOB.
DISPLAY “Enter the State to which the Child belongs”.
ACCEPT STATE.
*
EXEC SQL
INSERT INTO ORPHAN (NAME,FNAME,MNAME,DOB,STATE)
VALUES(:CNAME,:FAT-NAME,:M-NAME, :DOB,:STATE)
END-EXEC.
IF SQLCODE = 0
DISLAY “RECORD ADDED”
ELSE
DISPLAY ‘ERROR OCCURRED”.
STOP RUN.
*
0001-MAIN-PARA-EXIT.
EXIT.
 MULTIPLE-ROW TABLE MANIPULATION

Considering the following SELECT statement:

EXEC SQL
SELECT *
FROM TDEPT
WHERE DEPTNO LIKE ‘D%’
END-EXEC.

* This SELECT returns many rows from table TDEPT

* The rows are not ready for the application program to use because COBOL
manipulates data only one row at a time

* Manipulation of many rows of returning data from a DB2 database is done by

using the CURSOR facility.
 CURSOR FACILITY

* The CUROR facility allows a COBOL program to gain address ability to individual
row occurrences of a many-rows Result table.

* The following is the sequence for cursor processing.

 DECLARE CURSOR either in WORKING – STORAGE or ROCEUDRE DIVISION
 Move the values for any host variables used in the WHERE clause criteria used to
access the DB2 table
 OPEN the CURSOR (Do this only when the program is ready to use the data)
 FETCH rows until the ‘end of result table’ condition occurs
 CLOSE the CURSOR
 DECLARE CURSOR STATEMENT

* An SQL DECLARE statement

* Defines and associates a CURSOR with a SELECT statement and WHERE

clause criteria used to access a DB2 table

* Names the cursor and its specific SELECT which, when the cursor is
OPENed, will retrieve a Result Table whose rows satisfy the WHERE
clause in the DECLARE CURSOR statement.

FORMAT

EXEC SQL
DECLARE cursor-name CURSOR FOR
SELECT column name list
FROM data source(s)
WHERE conditions to be met (if any)
[FOR UPDATE OF column name list
END-EXEC.
 DECLARE CURSOR STATEMENT

EXAMPLE

EXEC SQL
DECLARE C1 CURSOR FOR
SELECT EMPNO,LASTNAME,JOBCODE
FROM EMPLOYEE
WHERE WORKDEPT=:HV-WORKDEPT
[FOR UPDATE OF JOBCODE]
END-EXEC.

The associated SELECT in this example returns all rows with WORKDEPT equal to
the value of the host variable HV-WORKDEPT;

The cursor C1 is a pointer to the Result Table of these rows

* The Result Table is accessed a row at a time through the FETCH

statement

* The FOR UPDATE OF clause must be specified to allow a row retrieved

through a cursor to be changed by the DML UPDATE statement, if
desire; the columns to be
updated must be named in the column list

* Multiple unique cursors can be DECLARED in a program.

 DECLARE CURSOR STATEMENT

* Besides the WHERE feature, the SELECT statement can utilize any or all of the
following SQL features:
 Built-in Functions
 HAVING clause
 GROUP BY clause
 ORDER BY clause

• If the SELECT within DECLARE CURSOR statement utilizes any of the following
features, the result table becomes ‘read-only’

 UNION operator
 DISTINCT keyword
 GROUP BY clause
 ORDR BY clause
 HAVING clause

• The Result Table will be read-only if the SELECT statement.

• References a read-only table/view
 Identifies more than one table/view
 Identifies a table/view accessed in a Sub SELECT
OPEN STATEMENT

* Opens the cursor

EXAMPLE

The following statement opens the cursor C1 previously declared and

associated with the SELECT that build a Result Table

EXEC SQL
OPEN C1
END-EXEC.

* When a cursor is open, DB2/SQL makes available the Result

Table associated with the cursor;

The cursor is positioned before the first row of the Result

Table and no
data is returned at this point
 FETCH STATEMENT

* Returns a row from the Result Table to the program’s data area
FORMAT

EXEC SQL
FETCH cursor-name
INTO host-variable
END-EXEC.

* Host-variable(s) correspond to the parameters in the associated cursor

statement

EXAMPLE

EXEC SQL
FETCH C1
INTO :HV-EMPNO,:HV-JCOD
END-EXEC.

* Advances the cursor position to the next row of the Result Table
making it
the current row;
DB2 then retrieves data from the current row into the program
variables
FETCH STATEMENT

*Advances the position of the current row pointer in the forward

direction only

Note No facility presently exists for moving the current pointer

backward, to the top, or bottom of the Result Table

* Can be executed on an OPEN cursor only;

If an error occurs during the processing, the cursor is CLOSED and an

error code
is returned

* If issued causing the cursor position to point after the last result table
row,
DB2 will set the SQLCODE field to + 100 and no data will be
transferred to the
host variables
 CLOSE STATEMENT

* Tells the DB2 system that the accessing of the Result Table is
completed

EXAMPLE

EXEC SQL
CLOSE C1
END-EXEC.

* The cursor can be CLOSED at any time but it gets automatically CLOSED
when:

 An Application Terminates

 A Program encounters an ABEND

 A Program issues a Commit or Rollback

* All OPEND cursors will be CLOSED at the program termination, but a CLOSE
statement is recommended for performance reasons.

* The CLOSE statement should be executed for the previously OPENED cursors
only
A SKELETAL CURSOR PROGRAM
SEGMENT
EXEC SQL
DECLARE C1 CURSOR FOR
SELECT EMPNO, JOBCODE
FROM EMPLOYEE
WHERE WORKDEPT=‘C01’
END-EXEC.

EXEC SQL
OPEN C1
END-EXEC.

EXEC SQL
FETCH C1
INTO :HV-EMPNO,:HV-JCOD
END-EXEC.

EXEC SQL
CLOSE C1
END-EXEC.

Intermediate Results
FETCH HV-EMPNO HV-JCOD
1 000030 60
2 000130 55
3 000140 56
A DB2/SQL COBOL PROGRAM WITH
CURSOR

Problem Definition
* To list the Name of the Child, its Date of Birth and the State to which
it belongs, for those orphans whose Father’s name is unknown
IDENTIFICATION DIVISION.
PROGRAM-ID. SAMPLE1.
*
ENVIRONMENT DIVISION.
*
DATA DIVISION.
WORKING – STORAGE SECTION.
01 WS-VARS.
05 CNAME PIC X(20).
05 FAT-NAME PIC X(20).
05 M-NAME PIC X(20).
05 DOB PIC 9(6).
05 STATE PIC X(10).
*
EXEC SQL
INCLUDE SQLCA
END-EXEC.
*
EXEC SQL
INCLUDE ORPHAN
END-EXEC.
A DB2/SQL COBOL PROGRAM WITH CURSOR

*
EXEC SQL
DECLARE NOPOP CURSOR FOR
SELECT NAME, DOB, STATE
FROM ORPHAN
WHERE FNAME =‘’
END-EXEC.
*
PROCEDURE DIVISION.
0001-MAIN-PARA.
EXEC SQL
OPEN CURSOR NOPOP
END-EXEC.
PERFORM 0011-CULL-PARA UNTIL SQLCODE = 100.
EXEC SQL
CLOSE NOPOP
END-EXEC.
STOP RUN.
 A DB2/SQL COBOL PROGRAM WITH CURSOR

0001-MAIN-PARA-EXIT.
EXIT.

0011-CULL-PARA
EXEC SQL
FETCH NOPOP
INTO :CNAME, :DOB, :STATE
END-EXEC.

IF SQLCODE NOT EQUAL TO 100

DISPLAYCNAME, DOB, STATE.
0011-CULL-PARA-EXIT.
EXIT.
Preparation Phase
 PREPARING A DB2/SQL COBOL PROGRAM FOR
EXECUTION

Steps involved in preparing a DB2/SQL program for execution (to translate

from the programmer’s environment into computer readable code)

* Precompile

* Compile & Link-Edit

* Bind
 PRE-COMPILER

DB2 precompile analyzes all of the embedded SQL statements.

 SQL statements are placed in a data base request module (DBRM).

 SQL statements are commented out in source and replaced with the host language
caLL

statement.

 DB2 precompile timestamps the DBRM modules and the program source code. This
timestamp is passed down to the program load module, and is also tied to the plan
by the DB2 catalog.

 A Data Base Request Module (DBRM) that will be used later when the program
goes to the BIND process is created
 BIND
Info about DB2 tables, indexes, etc.

*Has an its objective the creation of

an Application Access Plan in the DB2
environment. DB2 CATALOG

DBRM1
SYSIBM.SYSTABLESPACE
 *Bind
determines SYSIBM.SYSTABLES
SYSIBM.SYSCOLUMNS
access paths in the Extracted SQL statements SYSIBM.SYSINDEXES
from program SYSIBM.SYSCOLDIST
following ways:
D Data access strategy (plan)
 • Identifying which
indexes will be used
B
 • Defining sorting DBRM2 DB2 DIRECTORY

that will be performed DSNDB01.SCT02

 • Deciding what join

2 Information about bind

method will be used

Package List
 Access strategy is
DB2 CATALOG

determined at bind SYSIBM.SYSDBRM

SYSIBM.SYSPLAN
time. SYSIBM.SYSPLANAUTH
SYSIBM.SYSPLANDEP
SYSIBM.SYSPACKLIST
 Analyzing the DBRMs, it develops an SYSIBM.SYSPLSYSTEM
Application Access plan based on the SYSIBM.SYSSTMT
access SYSIBM.SYSPACKAUTH
requested and the database SYSIBM.SYSTABAUTH
components present. Optimizer

*This plan is retrieved and executed

every time the program is run
 BIND
Info about DB2 tables, indexes, etc.

COMPILE AND LINK-EDIT

DB2 CATALOG

* As the program is compiled, the DBRM1

SYSIBM.SYSTABLESPACE
various source modules become object SYSIBM.SYSTABLES
modules. Extracted SQL statements
SYSIBM.SYSCOLUMNS
SYSIBM.SYSINDEXES
from program SYSIBM.SYSCOLDIST
* These object modules are then link D Data access strategy (plan)
edited into load modules

BIND DBRM2
B DB2 DIRECTORY

DSNDB01.SCT02
*Has an its objective the creation of an
Application Access Plan in the DB2 2 Information about bind
environment.
Package List
*Analyzing the DBRMs, it develops an DB2 CATALOG
Application Access plan based on the SYSIBM.SYSDBRM
access SYSIBM.SYSPLAN
SYSIBM.SYSPLANAUTH
requested and the database SYSIBM.SYSPLANDEP
SYSIBM.SYSPACKLIST
components present. SYSIBM.SYSPLSYSTEM
SYSIBM.SYSSTMT
SYSIBM.SYSPACKAUTH
*This plan is retrieved and executed SYSIBM.SYSTABAUTH
Optimizer
every time the program is run
PREPARING A DB2/SQL COBOL PROGRAM FOR
EXECUTION

Precompiler * Is a preprocessor for the host programming languages

(PL/I,COBOL, etc.)
* Analyzes a host language source module, stripping out
all the SQL statements, it finds and replacing them by
host language CALL statements;
(At run time those Calls will pass control-indirectly- to
the Runtime Supervisor)
* Constructs with this information, a DBRM which
becomes
input to the Bind component
* Produces a source listing, showing the original source
code, diagnostics, cross-reference information, etc.

DBRM * Constructed by the Precompiler from the SQL statements

it encounters
* One per source module
* Stored a member of an MVS Partitioned Data Set
* Contains an edited form of the original SQL
statements,
together with certain additional information
 PREPARING A DB2/SQL COBOL PROGRAM FOR
EXECUTION

PACKAGE
* Produced by binding the DBRMs
* Consists of a set of internal structures, representing the compiled
from of the original SQL statements in the corresponding DBRMs
* Physically stored in the DB2 directory

COLLECTION

* A name given to a logically related set of packages

* Does not have nay physical existence of its own
* Typically (but not necessarily) all of the packages used in a given
application
would be assigned to the same collection

Packages and Collections

* Each package is assigned to exactly one collection when it is created

* Collections provide a useful “level of indirection”
 PREPARING A DB2/SQL COBOL PROGRAM FOR
EXECUTION

PLAN (Application Plan)

* Consists simply of a lit of the (names of the ) packages needed

to execute
that application

* Is produced by binding together the specified packages and/or

DBRMs

* The packages concerned are specified either by individual

package name or
by collection name (meaning, all packages in the specified
collection)
 PREPARING A DB2/SQL COBOL PROGRAM FOR
EXECUTION

* Is an optimizing compiler
* Converts high-level database requests (SQL statements) into optimized
internal
form
* Compiles SQL statements into “CODE” even though that code is not
true machine code per se

Code Set of internal control structures, that are used to drive a set of
generalized I/O routines within the Data Manager

* Has as its input, a DBRM and as its output, a package (the compiled
form of that DBRM)
* BINDs a given DBRM to produce a Package
* BINDs together a list of packages to produce an Application Plan or
simply plan

If a DBRM is thought of as a SQL source module, then the package produced

by binding that DBRM can be though of as the corresponding object
module

DB2 is thus a compiling system and Bind perform a compiling function for
SQL statements, much as the host language compiler provides a compiling
function for the host language statements in which those SQL statements are
embedded
PREPARING A DB2/SQL COBOL PROGRAM FOR
EXECUTION

Syntax Checking

* Bind examines the SQL statements in the input DBRM, parses them, and reports on
any syntax errors it finds.

Note Even though the Precompiler has already performed similar checks, as the
Precompiler is decoupled from the rest of DB2, Bind can’t assume that its input is a valid
precompiler output – the user might have constructed an invalid DBRM via some other
mechanism

The Precompiler can run even when DB2 is not available. It can even run on a
different machine – and its output is not automatically protected.

DataBase Request Validation

* Validates all the tables, views, and columns used in the program

* If any of the above objects is not present in the DB2 system, the BIND process will
terminate abnormally
 PREPARING A DB2/SQL COBOL PROGRAM FOR
EXECUTION

Optimization
* Bind includes an Optimizer as an important subcomponent
* The optimizer chooses, for each SQL statement it processes, an optional access
strategy for
implementing that statement
Note Data manipulation statements such as SELECT specify only what data the user,
wants, not how to get
to that data;
The access path for getting to that data will be chosen by the optimizer. Programs
are thus independent of such access paths.
* Even in a very simple case, there could be at least two ways of performing the
desired retrieval. The optimizer will choose one of the two available strategies to adopt
* In general, the optimizer will make its choice on the basis of such considerations as
the following:
- Which tables are referenced in the SQL statement?
- How big those tables are?
- How selective those indices are?
- How the data is physically clustered on the disk?
- The form of the data WHERE clause in the request and so on.
Optimization

* Bind will generate code that is tightly bound to the optimizer's choice of strategy

Eg If the optimizer decides to make use of an index called X, then the generated
package will include explicit references to index X

Q Where does the optimizer get its information from?

How does it know how big the tables are, or what indexes exist?

A This information is kept in the catalog, in the form of database statistics. A special
utility, RUNSTATS is provided for gathering such statistics and storing them in the
catalog for the optimizer’s use
Package Generation

* Is the actual process of building the package

Verifying Authorization

* Checks whether the user who is to be the owner of the bound package is
allowed

- to perform the operations requested in the DBRM to be bound and

- to assign packages to the applicable package collection

Plan Bind

* Has as its input, a list of packages and/or package collections and as its
output, the bound application plan

* Also does some authorization checking. Specifically, it checks that the user
who is to be

the owner of the bound plan is authorized to execute all of the applicable
packages
FINAL EXECUTION

* Since the original program has now effectively been broken into two
pieces (Load Module and Application Plan), those two pieces must
somehow be brought back together again at execution time
* First, the COBOL load module is loaded into memory; it starts to
execute in the usual way

* Sooner or later it reaches the first call to the DB2 Languages Interface
module. That module gets control and passes control in turn to the
Runtime Supervisor

*The Runtime Supervisor then retrieves the application plan (and

associated packages) from the DB2 directory, loads them into the
memory, and uses the information they contain to request the desired
function on the part of the Data Manager
*The Data Manager in turn performs the necessary operations on the
actual stored data and passes results back (as appropriate) to the COBOL
program
RUN TIME SUPERVISOR

* Is resident in main memory when the appln. Program is executing, to

oversee that execution
* Is the firs to get the control, when the appln. Program requests some
database operation to be performed
* Uses the control information to request the appropriate operations on the
part of the Data Manager.

DATA MANAGER
* Is the component that manages the physical databases by performing all of
the normal access method functions like search, retrieval, update, index
maintenance.
* Invites other system components as necessary in order to perform detailed
functions as locking, logging, I/O operations, etc. during the performance of
its basic task

BUFFER MANAGER
* Is the component responsible for physically s data between external storage
and memory
* Employs sophisticated techniques such as read-ahead buffering and look-
aside buffering to get the best performance out of the buffer pools under its care
and to minimize the amount of physical I/O actually performed.