Transcript Modeling_Patterns

Data Modeling on Steroids: Patterns
and Reusability
Presented by Bert Scalzo, PhD
[email protected]
About the Author






Oracle DBA for 18+ years, versions 4 through 10g
Worked for Oracle Education & Consulting
Holds several Oracle Masters (DBA & CASE)
BS, MS, PhD in Computer Science and also an MBA
LOMA insurance industry designations: FLMI and ACS
Books
– The TOAD Handbook (March 2003)
– Oracle DBA Guide to Data Warehousing and Star Schemas (June 2003)
– TOAD Pocket Reference 2nd Edition (June 2005)
 Articles
– Oracle Magazine
– Oracle Technology Network (OTN)
– Oracle Informant
– PC Week (now E-Magazine)
– Linux Journal
– www.Linux.com
About Quest Software
We provide application management
solutions that enable customers to:
•Develop
•Deploy
•Manage
•Maintain
enterprise applications without downtime
What is Application Management?
A holistic approach of managing the entire
application - not individual components
How Do We Do It?
We surround the application infrastructure:
Application Monitoring
Database Management
High Availability
Microsoft Infrastructure Management
Agenda




Definition of Pattern
Benefits of Designing for Reusability
Super and Sub Type Pattern
Intersection Patterns
– Many to many relationships
– Roles
– Assemblies
World of Modeling …
• End-user
• IT Partner/Liaison Business Process Modeling
(BPM)
• Business Analyst
• Improve process efficiency
• Define/document Bus. Processes
- create correct and complete
application requirements
• System Architect
• System Analyst
• App Developer
• Support for all UML diagrams
- Analyze requirements
- Design application
• Reverse/forward engineer code
• Bus. Analyst
• Data Architect
• Data Analyst
• DBA
• DB Developer
• DB Architect
Object-Oriented Modeling
(OOM - UML)
Conceptual Data Modeling
(CDM – E/R)
Physical Data Modeling
(PDM)
QDesigner synchronizes models
from all levels in a single tool
• Identify all data & relationships
- E/R (Entity/Rel’ship) diagrams
- DB independent view
• Business Rules?
•
•
•
•
DB-specific model
Reverse engineer existing DB
Create/Update DB from model
Data Warehouse Modeling
Basic Premises of Design Patterns
 Not everything is specialized
 Simply reuse what’s reusable
 Anyone can do this (really…)
 Not that hard to do technically
 Merely requires some up front work
Not Everything is Custom
 How many of your businesses are so
specialized that you require custom
modeling of everything?
 We all have:
–
–
–
–
–
–
Customers
Products
Contracts
Orders
Addresses
etc, etc, etc …
These base concepts are
often repeated within one
company across projects,
departments, divisions …
Side Effects of Customization
 I had one insurance company for:
–
–
–
–
–
–
Life and Accident
Automobile
Home Owners
Health Insurance
Administer 401K
Variable Annuity
Each statement was mailed
using a different address style
(i.e. diff info) , and it took 6
separate change of address
cards when I moved. So I
switched insurance companies
 I later worked for that company
– Nearly 1000 entities
– Close to 20,000 attributes
– In a year, we simplified the model by 90%!!!
Benefits of Re-Usable Models
Takes time, but often yields:
 Better data models
– More robust fundamental business knowledge
 Better database design
– More accurate implementation of business requirements
 Less rework
– Avoids last minute implementation/coding discoveries
 Less end-user re-training
– Systems won’t need to be reworked and redeployed
 Centralized databases
– All can be based upon common modeling best practices
Anyone Can Re-Use Models
Regardless of “IS” shop organization
 Centralized Modeling group – easy
 Centralized Data Admin – again, easy
 Departmental or Project Oriented:
– Merely requires open communication 
– Helps to all use the same modeling tool
– Helps for everyone to follow some fundamental
modeling rules, guidelines and best practices
– Can still work in a limited fashion with none of the
above – through grass roots efforts to work smarter
 Consultants – simply require all data models
be part of walk-throughs and deliverables
(help them learn the business)
Simple Pattern: 2 Constructs

Simply requires the appropriate,
consistent and liberal use of just two very
basic data modeling constructs
1. Super and Sub Types
2. Bridging/Intersection Entities
– NOTE: There are two highly specialized intersection
scenarios for making reusable modeling components
• Roles
• Assemblies (sometimes called B.O.M. solution)
Super and Sub Types
 In the business world, we often find that things (i.e.
entities) do not fall into simple, clear cut classifications
– We sometimes find that things are very similar - with just a few
differences
 In data modeling, this can be referred to as:
– Entity Inheritance
– Super and Sub Types
– Generalization Hierarchies
 The key question to ask is whether these different
classifications are:
– Exclusive: must be one way or the other
– Inclusive: can actually be all ways at once
http://www.quest.com/whitepapers/Reality_Requires-QDesigner.pdf
Candidate for Super/Sub
Very similar entities in
concept - some common
attributes, but also some
differing attributes …
Super/Sub Type Pattern
X means Exclusive
Super/Sub
Meta-Data
This Super/Sub Type Meta-Data is
only important for transforming
conceptual to physical model…
Intersection Patterns

There are three ways to use intersections:
1. Resolve basic many-to-many relationships
• Not very intrinsically useful, but must mention
for clarity
2. As “Roles”
• Powerful concept
• Can greatly reduce data model size
3. As “Assemblies”
• Powerful concept
• Can enable flexible database design
Resolve Many-To-Many
 You CANNOT physically implement the concept of many-tomany relationships in a relational database
 You may potentially miss multiple key business requirements
 Many modeling tools will attempt to automatically resolve this
Intersection Meta-Data
Candidate for Roles
Unlike Super/Sub types, we now have identical or near
identical entities coming together into an intersection …
Poor Choice for Roles

Often, a first attempt to resolve the “Role” concept uses
relationships. But there are two problems with this approach:
1. What if “Role” participation requires additional attributes?
Where would you place them?
2. What if business requirements change and you need to
add or subtract “Role” participants? How do you later add
and subtract foreign keys in databases?
Solution for Roles
By creating a “Role Type” as entity rather than attribute,
adding/dropping Roles doesn’t require database change!
Candidate for Assemblies
Solution for Assemblies
This does everything the prior model does, with the advantage
of not being fixed in terms of what things can be combined or
in levels of hierarchical combinations possible …
Parts
Minor Components
Major Components
Products
Assemblies Good for Anything
 At insurance company, we used Assemblies to build
dynamic product offerings comprised of:
– Features
– Federal Rules
– State Rules
– etc, etc, etc …
 Real Benefit: new products required no new application
systems, just a few new rows of data in existing tables
 Product time to market reduced by 50%!!!
Assemblies Work …
Component
Type
Component
Assembly
1
Part
1
Screw
1
5
1
2
Minor
2
Nut
1
5
2
3
Major
3
Washer
1
5
3
4
Product
4
Bracket
1
6
1
5
Widget
2
6
4
6
Gadget
2
7
5
7
Sprocket
3
7
6
8
Cog
3
8
1
9
Gizmo
4
8
5
9
7
9
8
How many base parts in a Gizmo? (nine)
Did You Notice …???
Roles and Assemblies are
nearly the same thing …
Parting Thoughts
 Many Design problems have already been solved
 Data Model constructs can be re-used as easily as code
– Carefully consider which business requirements are really unique
– Must look for ways to reduce model complexity while making it “generic”
 There are many other modeling issues – this was just a start …
– Breaking models into sub-models
– Round-trip Engineering:
• Conceptual -> Physical Model compare and sync
• Physical Model -> Database compare and sync
– Repository-based collaborative modeling
– Horizontal and Vertical Partitioning
– Data Warehousing (Star Schema design)
– Object-Relational Mapping
– etc, etc, etc …
More Information
 Free 45-Day Trial of QDesigner™
– http://www.quest.com/QDesigner
– Compete enterprise modeling solution
• Business Process Modeling (BPM)
• Object-Oriented (UML) Modeling (OOM)
• Conceptual Data Modeling (CDM)
• Physical Data Modeling (PDM)
 Save $$$: Upgrade your TOAD/SQL Nav license to the
Suite
– Suites include PDM!!!
 Modeling White Papers
– http://www.quest.com/whitepapers/qdesigner/index.asp
• Data Modeling: Common Mistakes and Their Impact
• Data Modeling: It's Really All About the Relationships
• Data Modeling: Reality Requires Super and Sub Types
QUESTIONS & ANSWERS