IBM Systems Lab Services and Training A Data Centric ...

IBM Systems Lab Services and Training

A Data Centric Approach
Development - Using SQ
Lifting
Quser January 15, 2013

Dan Cruikshank
[email protected]

[email protected]
ibm.com/systems/services/labservices

h to Application
QL to do the Heavy
3

© 2011 IBM Corporation

IBM Systems Lab Services and Training

Agenda

 Quser Back in the Day
 Data Centric Development
 Transparent Migration from DDS to D

2 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

DDL © 2011 IBM Corporation

ices

IBM Systems Lab Services and Training

Quser Back in the Day

 I served as an officer of Quser for 4 y
–I was preceded by Dick Jacobson

 Notable Speakers
–Ken Kelley President and founder
• Makers of Abstract/Probe and S
• Topic: Object Oriented Program
–Paul Conte Author and DB2 for i D
• First professional speaking eng
• Topic: Relational Database De

 Implicit/Explicit Record Format sharin
–Foundation for transparent DDS to
–Concepts were introduced at Quse
errors without recompiling

3 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

years from 1983-86
founder of Help/Systems

of Advanced System Concepts
SEQUEL (now part of Help/Systems)
mming
Database Guru
gagement
esign

ng
o DDL migration
er as technique to avoid level check

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Data Centric Development

 Data Centric Programming and Set B
 Defining Sets using SQL DDL
 Accessing Sets using SQL DML

4 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

Based Processing

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

What is Data Centric Programming

 Definition:
–Solving business problems using
–Getting the database managemen
behalf
–Implementing more of the busines
–Separating the business logic from
programs

5 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

g?

database capabilities
nt system to do more on your
ss logic in the database
m the high level application

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Application Centric Overview
Resu

HLL PGM

DDS

T1 T2

6 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ults
M

T3 T4

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Application Centric Example

 More application logic
 Small pipe (1 row at a time)
 Not using much intelligence in the

High Level Language R
Program

Establish begin and
end dates

READ File
Convert numeric date

to real date
If real date is => begin date
And real date <= end date

Then
Do some work

End If

7 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

database

READ Data DDS Database

Numeric_Date

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Data Centric Overview

HLL PGM

Resu

DB2

T1 T2

8 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

M or Interface
ults

T3 T4

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Data Centric Programming Exampl

 Less application logic

– Less application developme
– You can run and test SQL w

 Big pipe (sets of many rows proce
 Taking advantage of more intellige

HLL Program FETCH Data

Do some work

9 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

le

ent and testing
without any programming

essed at a time)
ence in the database

DB2 Database

SELECT *
FROM View
WHERE real_date
BETWEEN Begin_date
AND End_date

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Data Centric Programming Scalabi

 Goal is to drive as much work down into
the database management system as
possible.

 Define business rules as part of the
database
– Rules apply to all application
interfaces

 Take advantage of SQL only
capabilities

 Database evolves to:
– Meet new requirements
– Take advantage of new technology

10 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ility

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

SQL and Data Centric Programmin

SQL works very well
Data Centric Program

SQL
sets

11 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ng

for
mming

works best for processing
of data

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Set Based Thinking and SQL

 SQL is declarative and not procedura
–Think in terms of defining the sets

 DDL (Data Definition Language)
–Defines the sets

 DML (Data Manipulation Language)
–Operates upon those sets

12 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

al
s and the operations on those sets

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Enabling Sets

Data Centric DBMS kno
ve

Program kno

13 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ows about data
ersus
ows about data Traditional

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

SQL and Data Centric Programmin

Proper data modeling re
in good “set” definitio

Proper
good

14 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ng

esults
ons

r SQL usage results in
set based operations

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Agenda

 Data Centric Programming and Set
 Accessing Sets using SQL DML

15 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

Based Processing

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Data Centric Constructs Used Whe

 Constraints
–The basis for a solid foundation

 Indexes
–The plumbing and wiring

 SQL Views
–The walls, roof, doors and window

 DB2 for i Auto Generated Column Su
–The “must have” appliances

16 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

en Defining the Database

ws © 2011 IBM Corporation
upport

ices

IBM Systems Lab Services and Training

What are the Data Integrity Rules?

Customers Ord
CustomerNo OrderNo
Name LineNo
Address Custom
State ItemNo
City Quantity
Zip Price

What happens if we are updating Orders but hav
What happens if we drop a customer row from C
What happens if we add an element to Items wit

17 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ders Items
o ItemNo
Description
merNo Color
Size
y

ve specified an invalid CustomerNo value?
Customers?
th an invalid size?

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Constraints

 Unique key constraint
– the values of the key are valid only if they
– More than 1 per table

 Primary key constraint
– the values of the key are valid only if they
– Only 1 per table

 Referential constraint
– the values of the “foreign key” are valid on
• foreign key appear as values of a pare
• the foreign key is null

 Check constraint
– the rule that limits the values allowed in a

18 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

are unique

are unique and not null-able

nly if one of the following conditions is met:
ent key

column or group of columns

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Enforcing the Rules

Customers Ord
CustomerNo OrderNo
Name LineNo
Address Custom
State Item_ID
City Quantity
Zip Price
Customer_ID
Constraints
All UNIQUE
applications PRIMARY KEY
must follow FOREIGN KEY
the rules
CHECK

19 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ders Items
o ItemNo
Description
mer_ID Color
D Size
y Item_ID

Rules

Identifies unique public key

Y Identifies parent key

Y Matches parent key. Defines
actions for insert, update and
delete events

Ensures proper values, e.g.
Quantity > 0

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Indexes

 Indexes are needed for optimal perfo
–1 Index can service several SQL r

 Indexes are not referenced directly in
UPDATE, etc.
–That’s the job of the query optimiz
use indexes

 Indexes can contain:
–Expressions: CREATE INDEX X1
–Functions: CREATE INDEX X2 ON
–Aggregates: CREATE EVI E1 ON
–Non-keyed columns: CREATE IND
COLUMNS ID, NAME
–Row filtering: CREATE INDEX X4
‘D11’

20 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ormance
requests
n SQL statements like SELECT,

zer, it decides if and when and how to

ON T1 (QTY * PRICE)
N T1 (UPPER(NAME))
T1 (NAME) INCLUDE(SUM(QTY))
DEX X3 ON T1 (DEPT) ADD

4 ON T1 (NAME) WHERE DEPT =

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

INDEX Usage Examples

SELECT …. FROM T1 ORDER BY DEPT;
SELECT …. FROM T1 WHERE DEPT =

‘D11’ ORDER BY NAME;

SELECT …. FROM T1 WHERE DEPT =
‘D11’ AND SUBSTR(NAME,1,3) = ‘ABC’
ORDER BY NAME;

SELECT DEPT, SUM(QTY) FROM T1
GROUP BY DEPT
ORDER BY DEPT;

21 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

One index servicing two queries

CREATE INDEX X3 ON T1 (DEPT, NAME);
Multiple Indexes servicing 1 query

CREATE INDEX X3 ON T1 (DEPT, NAME);

CREATE INDEX X7 ON T1 (SUBSTR(NAME,1,3));

An aggregate Index supporting SUM
CREATE INDEX E7 ON T1 (DEPT)
INCLUDE SUM(QTY);

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

SQL Views

 SQL views are used to show some su
isolate some complex operation

 SQL views do not contain data or acc
 Views have no implied order
 Views are referenced just like tables
 Views can be joined to other views
 Views can be used to externally desc

22 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

ubset of the data or to capture and
cess paths

in SQL statements

cribe data structures © 2011 IBM Corporation

ices

IBM Systems Lab Services and Training

VIEW Usage Examples

Complex Query

SELECT … FROM T1, T2, T3,T4…
GROUP BY… ORDER BY…

Program Described Structure

T1_RESULT DS
A CHAR(10)
B DECIMAL(11,0)
C DATE

SELECT A, B, C INTO :T1_RESULT FROM T1

Multiple Views and Data Structures

T1_RESULT DS EXTNAME T1_VIEW
T2_RESULT DS EXTNAME T2_VIEW

SELECT * INTO :T1_RESULT FROM T1_VIEW
SELECT * INTO :T2_RESULT FROM T2_VIEW
WHERE T2.A = :T1_RESULT.A

23 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

Masking Complexity

CREATE VIEW COMPLEX_QUERY
AS SELECT … FROM T1, T2, T3,T4… GROUP BY…;

SELECT * FROM COMPLEX_QUERY ORDER BY…;

Externally Described Structure

CREATE VIEW T1_VIEW
AS SELECT A, B, C FROM T1;

T1_RESULT DS EXTNAME(T1_VIEW)

SELECT * INTO :T1_RESULT FROM T1_VIEW

Joined View One Structure

CREATE VIEW JOIN_VIEW AS
SELECT A, B, C, D, E, F FROM T1 JOIN T2 USING(A);

JOIN_RESULT DS EXTNAME(JOIN_VIEW)
SELECT * FROM JOIN_VIEW INTO :JOIN_RESULT

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Auto-generated Values

 DB2 for i can automatically generate
 A value can be auto-generated in 1 o

–Defined as a column attribute
–Defined as a column type (ROWID
–Extracted from a external object
 The following SQL column attributes
–Row Change Timestamp
– Identity
 A Sequence object contains a system
–Is external from a table
–Can be used on INSERT statemen

24 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

the value for a column
of following ways:
D)

allow auto-generation:

m generated value
nts to assign the value to a column

ices © 2011 IBM Corporation

IBM Systems Lab Services and Training

Auto-Generated Column Exampl

Row Change Timestamp

CREATE TABLE ORDERS
(ORDER_KEY BIGINT …,
ORDER_CHANGE_AUDIT TIMESTAMP

NOT NULL
FOR EACH ROW ON UPDATE AS ROW

CHANGE TIMESTAMP)
IMPLICITLY HIDDEN;

Identity Column

CREATE TABLE ORDERS
(ORDER_ID BIGINT NOT NULL
GENERATED BY DEFAULT AS IDENTITY

PRIMARY KEY IMPLICITLY HIDDEN,

25 IBM Systems Lab Services and Training – ibm.com/systems/services/labservi

les

Notes

IMPLICITLY HIDDEN prevents column from
being included in SQL * statements

ROW CHANGE TIMESTAMP function can be
used without column name knowledge

SELECT …. ROW CHANGE TIMESTAMP FOR
T1 AS RCTS FROM T1

UPDATE T1 ….. WHERE ORDER_KEY =
:v_ORDER_KEY AND ROW CHANGE
TIMESTAMP FOR T1 = :v_RCTS

ID Column Notes

Only 1 IDENTITY column per table

Ideal as meaningless PRIMARY KEY

GENERATED ALWAYS results in the
generation of a new value automatically.

Existing values are ignored
Can be an issue when copying or duplicating data

GENERATED BY DEFAULT only generates a
value if the auto-generated column is not
included in the data

ices © 2011 IBM Corporation