Data Modelling Contentious Issues The Best Debates from the Data Modelling List January 2005 Karen Lopez InfoAdvisors.com The Best Debates from the Data

Data Modelling Contentious Issues

Data Modelling Contentious Issues

The Best Debates from the Data Modelling List

January 2005

Karen LopezInfoAdvisors.com

The Best Debates from the Data Modelling List

January 2005

Karen LopezInfoAdvisors.com

7 Mar 01Jan 20057 Mar 01Jan 2005

© 2001 InfoAdvisors © 2001 InfoAdvisors 22

Karen López, I.S.P.Karen López, I.S.P.

• Karen is the moderator of the Data Modelling List. She has 18 years of data modelling experience on large, multi-project programs

• She has a B.Sc. in Computer Technology / Information Systems from Purdue University

• She is a former President of the Information Resource Management Association of Canada (IRMAC)

• Karen is the moderator of the Data Modelling List. She has 18 years of data modelling experience on large, multi-project programs

• She has a B.Sc. in Computer Technology / Information Systems from Purdue University

• She is a former President of the Information Resource Management Association of Canada (IRMAC)

7 Mar 01Jan 20057 Mar 01Jan 2005


Upcoming Speaking Engagements

Upcoming Speaking Engagements

• DAMA / Meta Data SymposiumMay 2005

• Toronto IRMACSpring 2005

• DAMA / Meta Data SymposiumMay 2005

• Toronto IRMACSpring 2005

7 Mar 01Jan 20057 Mar 01Jan 2005


About this PresentationAbout this Presentation

• We will be using an interactive format - you will be participating in informal polls about data modelling issues and best practices.

• This is not an introductory presentation - a good knowledge of data modelling issues will be assumed.

• We will be using an interactive format - you will be participating in informal polls about data modelling issues and best practices.

• This is not an introductory presentation - a good knowledge of data modelling issues will be assumed.

7 Mar 01Jan 20057 Mar 01Jan 2005


InfoAdvisors’ Discussion Groups

InfoAdvisors’ Discussion Groups

• E-mail, web, and newsgroup based discussion group

• Data Modeling, Frameworks, Tools Groups

• Over 8000 subscribers • Moderated• No Charge• www.infoadvisors.com

• E-mail, web, and newsgroup based discussion group

• Data Modeling, Frameworks, Tools Groups

• Over 8000 subscribers • Moderated• No Charge• www.infoadvisors.com

http://www.infoadvisors.com/


7 Mar 01Jan 20057 Mar 01Jan 2005


E-mailE-mail

7 Mar 01Jan 20057 Mar 01Jan 2005


WebWeb

7 Mar 01Jan 20057 Mar 01Jan 2005


NewsgroupsNewsgroups

7 Mar 01Jan 20057 Mar 01Jan 2005


AgendaAgenda

• Contentious Issues• Background• Discussion Quotes• Poll• Results & Analysis

• Resources

• Contentious Issues• Background• Discussion Quotes• Poll• Results & Analysis

• Resources

7 Mar 01Jan 20057 Mar 01Jan 2005


Contentious IssuesContentious Issues

Near-religious discussions and debates

– People rarely change their minds based on this discussion

– The most successful discussions are ones where both sides learn something new about the other viewpoint.

Near-religious discussions and debates

– People rarely change their minds based on this discussion

– The most successful discussions are ones where both sides learn something new about the other viewpoint.

7 Mar 01Jan 20057 Mar 01Jan 2005


Conceptual Data ModelsConceptual Data Models

• Do you do them?• Are they used? How?• Just what is one?• How do they differ from other

Data Models?

• Do you do them?• Are they used? How?• Just what is one?• How do they differ from other

Data Models?

7 Mar 01Jan 20057 Mar 01Jan 2005



What is a CDM?– Entity only, no cardinality, fewer than 15

entities– Entity only, cardinality, more than 15

entities– Entity only, one Entity for every entity in

the LDM– Entity & Attributes, no physical, no

surrogate keys– Entity & Attributes, no DBMS-specific

issues

What is a CDM?– Entity only, no cardinality, fewer than 15

entities– Entity only, cardinality, more than 15

entities– Entity only, one Entity for every entity in

the LDM– Entity & Attributes, no physical, no

surrogate keys– Entity & Attributes, no DBMS-specific

issues

7 Mar 01Jan 20057 Mar 01Jan 2005


Conceptual Data Models – View 1


The conceptual model is concerned with the real world view and understanding of data; the logical model is a generalized formal structure in the rules of information science; the physical model specifies how this will be executed in a particular DBMS instance.

Duncan Dwelle, AIS Intl.http://www.aisintl.com/case/CDM-PDM.html

The conceptual model is concerned with the real world view and understanding of data; the logical model is a generalized formal structure in the rules of information science; the physical model specifies how this will be executed in a particular DBMS instance.

Duncan Dwelle, AIS Intl.http://www.aisintl.com/case/CDM-PDM.html

http://www.aisintl.com/case/CDM-PDM.html

http://www.aisintl.com/case/CDM-PDM.html

7 Mar 01Jan 20057 Mar 01Jan 2005




“A [CDM], typically called an Entity Relationship Diagram (ERD), contains business information and structure without regard to physical storage concerns. Business items become entities and you describe them with attributes. You formalize relationships…. Typically you have no index, foreign key or tablespace information. …A Logical Data Model (LDM) is sometimes skipped over. It is a hybrid between the CDM and PDM and contains some DBMS elements like denormalization and indexes, but isn't as detailed as the PDM.”

- Michael N.

“A [CDM], typically called an Entity Relationship Diagram (ERD), contains business information and structure without regard to physical storage concerns. Business items become entities and you describe them with attributes. You formalize relationships…. Typically you have no index, foreign key or tablespace information. …A Logical Data Model (LDM) is sometimes skipped over. It is a hybrid between the CDM and PDM and contains some DBMS elements like denormalization and indexes, but isn't as detailed as the PDM.”

- Michael N.

7 Mar 01Jan 20057 Mar 01Jan 2005




“[CDM is] a tricky phrase to define in that our industry uses this same phrase for two very different (in my opinion) ways. If you come from the Information Engineering background, a CDM is an entity and relationship-only method (no attributes and often no cardinalities to the relationships) that has a purpose of defining the scope of smaller, more detailed logical data modelling projects. Those logical data models are not constrained by organizational, platform, or other technical issues.

If you come from an ORM background, CDM is much like Mike has described [previously]. As he mentions, in this case Logical Data Modelling is similar to a first cut physical data model.

- Karen Lopez

“[CDM is] a tricky phrase to define in that our industry uses this same phrase for two very different (in my opinion) ways. If you come from the Information Engineering background, a CDM is an entity and relationship-only method (no attributes and often no cardinalities to the relationships) that has a purpose of defining the scope of smaller, more detailed logical data modelling projects. Those logical data models are not constrained by organizational, platform, or other technical issues.

If you come from an ORM background, CDM is much like Mike has described [previously]. As he mentions, in this case Logical Data Modelling is similar to a first cut physical data model.

- Karen Lopez

7 Mar 01Jan 20057 Mar 01Jan 2005


Conceptual Data Models – Warm up Vote

Conceptual Data Models – Warm up Vote

• Do Conceptual Data Models have Attributes?

• Do they have cardinality?• Are there fewer Entities in a

CDM than in an LDM?• If they have attributes, can they

have surrogate keys?

• Do Conceptual Data Models have Attributes?

• Do they have cardinality?• Are there fewer Entities in a

CDM than in an LDM?• If they have attributes, can they

have surrogate keys?

7 Mar 01Jan 20057 Mar 01Jan 2005


Vote!Vote!

Ok, Here’s How we vote1. I post a question2. The question has a Range.3. You put the sticky in the area for

your vote (1-5), sometimes once for PDM and once for LDM.

4. We debate the answer and the results

Ok, Here’s How we vote1. I post a question2. The question has a Range.3. You put the sticky in the area for

your vote (1-5), sometimes once for PDM and once for LDM.

4. We debate the answer and the results

7 Mar 01Jan 20057 Mar 01Jan 2005


Conceptual Data Models - Vote

Conceptual Data Models - Vote

Do you create Conceptual Data Models …that are actively used?

1. Yes, it’s up-to-date, easily available, and used at least a couple of times a year

2.

3.

4.

5. What’s a Conceptual Data Model? We don’t need no stinkin’ CDMs!

Do you create Conceptual Data Models …that are actively used?

1. Yes, it’s up-to-date, easily available, and used at least a couple of times a year

2.

3.

4.

5. What’s a Conceptual Data Model? We don’t need no stinkin’ CDMs!

7 Mar 01Jan 20057 Mar 01Jan 2005



TRAP!You may find yourself debating a topic because there is no common definition of a model or object, not because there is a real disagreement.

TRAP!You may find yourself debating a topic because there is no common definition of a model or object, not because there is a real disagreement.

7 Mar 01Jan 20057 Mar 01Jan 2005


Do we need Classwords?Do we need Classwords?

• A traditional naming convention• Usually means there’s a

standard classword list• Some tools can check for

standards compliance• Examples: Date, Amount, Count,

Quantity

• A traditional naming convention• Usually means there’s a

standard classword list• Some tools can check for

standards compliance• Examples: Date, Amount, Count,

Quantity

7 Mar 01Jan 20057 Mar 01Jan 2005


Attribute NamesAttribute Names

• Customer First Name• Customer

• Backordered Specialty Order Item ID

• Item ID• Item

• Customer First Name• Customer

• Backordered Specialty Order Item ID

• Item ID• Item

7 Mar 01Jan 20057 Mar 01Jan 2005


Classwords - VoteClasswords - Vote

Does a good LDM / PDM data model use classwords?

1. Always2. 3. 4. 5. Classwords are so obsolete….

Does a good LDM / PDM data model use classwords?

1. Always2. 3. 4. 5. Classwords are so obsolete….

7 Mar 01Jan 20057 Mar 01Jan 2005


Keys - Natural or Surrogate

Keys - Natural or Surrogate

• Just what you call them may show your true colours....– Surrogate, Non-intelligent,

Unnatural, Dataless or Meaningless– Natural, Intelligent, Normal,

Cluttered

• Just what you call them may show your true colours....– Surrogate, Non-intelligent,

Unnatural, Dataless or Meaningless– Natural, Intelligent, Normal,

Cluttered

7 Mar 01Jan 20057 Mar 01Jan 2005


Keys – Natural or Surrogate

Keys – Natural or Surrogate

Order Number--------------------------

Order DateSales Person ID

Order NumberOrder Line Number--------------------------Product Number

Quantity

Order Number--------------------------

Order DateSales Person ID

Order Detail Sys Number--------------------------Order Line Number

Product NumberQuantity

7 Mar 01Jan 20057 Mar 01Jan 2005


Keys - Natural or Surrogate – View 1Keys - Natural or

Surrogate – View 1I have inherited many [tables] that have used business elements for keys. The complexity of the keys for some tables is horrendous…Another thing I hate about this method is putting business information into the PK. This leads to the business changing either the contents of, or the definition of, a key column. …You should never change PK values….

With single column PK, you have single column FKs and easier maintenance of the system over time. You may have to perform more joins to satisfy complexqueries, but the joins are simpler. … I have implemented surrogate keys as a "strongly recommended“ guideline for all future database development.

- Michael N.

I have inherited many [tables] that have used business elements for keys. The complexity of the keys for some tables is horrendous…Another thing I hate about this method is putting business information into the PK. This leads to the business changing either the contents of, or the definition of, a key column. …You should never change PK values….

With single column PK, you have single column FKs and easier maintenance of the system over time. You may have to perform more joins to satisfy complexqueries, but the joins are simpler. … I have implemented surrogate keys as a "strongly recommended“ guideline for all future database development.

- Michael N.

7 Mar 01Jan 20057 Mar 01Jan 2005


Keys - Natural or Surrogate – View 2Keys - Natural or

Surrogate – View 2I agree wholeheartedly with Mike, but with one exception. I think reference tables would be better retaining the business elements in the primary key since they're usually standalone tables not used in joins. On a rather large application we set the ref tables up with sequence numbers as the primarykey. Quickly some developers began hard coding them to refer back to the row. This caused us much difficulty

- Thomas Z.

I agree wholeheartedly with Mike, but with one exception. I think reference tables would be better retaining the business elements in the primary key since they're usually standalone tables not used in joins. On a rather large application we set the ref tables up with sequence numbers as the primarykey. Quickly some developers began hard coding them to refer back to the row. This caused us much difficulty

- Thomas Z.

7 Mar 01Jan 20057 Mar 01Jan 2005


Keys - Natural or Surrogate - VoteKeys - Natural or Surrogate - Vote

What best describes your approach in an LDM / PDM?1. Surrogate Keys? We don’t need surrogate keys - we have natural identifiers.

2.

3.

4.

5. Every entity deserves its own surrogate key.

What best describes your approach in an LDM / PDM?1. Surrogate Keys? We don’t need surrogate keys - we have natural identifiers.

2.

3.

4.

5. Every entity deserves its own surrogate key.

7 Mar 01Jan 20057 Mar 01Jan 2005


Derived and Redundant Data Dilemma

Derived and Redundant Data Dilemma

• Not all derived data is calculated• Snapshot versus History

– History: Just maintain a foreign key (or relationship) back to the timestamped data

– Snapshot: Copy the data to another location to preserve the data at that point in time.

• Not all derived data is calculated• Snapshot versus History

– History: Just maintain a foreign key (or relationship) back to the timestamped data

– Snapshot: Copy the data to another location to preserve the data at that point in time.

7 Mar 01Jan 20057 Mar 01Jan 2005


• Historical (no derived data)

• Derived and

• Snapshot

• Historical (no derived data)

• Derived and

• Snapshot

Derived Data DilemmaDerived Data Dilemma

Product NumberEffective Date-----------------Product Price

Order ID________Order Date

Order IDOrder Line ID----------------Product NumberQuantity

Product Number-----------------Description

Product NumberEffective Date-----------------Product Price

Order ID----------Order DateOrder Total Amount

Product Number-----------------DescriptionPrice

Order IDOrder Line ID----------------Product NumberQuantityProduct PriceOrder Line Total

7 Mar 01Jan 20057 Mar 01Jan 2005


Derived Data Dilemma - Vote

Derived Data Dilemma - Vote

Do Derived Attributes belong in an LDM/PDM?1. Never2. 3. 4. 5. All the time - pure logical data

modelling doesn’t work in the real world.

Do Derived Attributes belong in an LDM/PDM?1. Never2. 3. 4. 5. All the time - pure logical data

modelling doesn’t work in the real world.

7 Mar 01Jan 20057 Mar 01Jan 2005


Derived Data DilemmaDerived Data Dilemma

TIP!Every decision is ultimately made based on cost, benefit, and risk. Be prepared to analyze all three.

TIP!Every decision is ultimately made based on cost, benefit, and risk. Be prepared to analyze all three.

7 Mar 01Jan 20057 Mar 01Jan 2005


Abstraction / GeneralizationAbstraction /

Generalization• A modelling design decision• Can be very flexible• Can be very difficult to

understand

• A modelling design decision• Can be very flexible• Can be very difficult to

understand

7 Mar 01Jan 20057 Mar 01Jan 2005


AbstractionAbstraction

MEASUREMENT

VOLUMEMEASUREMENT

TEMPERATUREMEASUREMENT

WEIGHTMEASUREMENT

MEASUREMENT TYPE

7 Mar 01Jan 20057 Mar 01Jan 2005


Abstract or Specific - View

Abstract or Specific - View

“Although your model represents a higher level of abstraction than most people would create it does not violate the rules of normalization. So, it is not technically speaking incorrect.

Now come the issues. What is the purpose of the model? Are you documenting the business requirements and rules? If so your approach does not provide as much information about the business as a less abstract approach would…If you are trying to model a flexible environment where the required data (represented by the repeating groups) changes frequently then your model would be more appropriate.”

- Rick B.

“Although your model represents a higher level of abstraction than most people would create it does not violate the rules of normalization. So, it is not technically speaking incorrect.

Now come the issues. What is the purpose of the model? Are you documenting the business requirements and rules? If so your approach does not provide as much information about the business as a less abstract approach would…If you are trying to model a flexible environment where the required data (represented by the repeating groups) changes frequently then your model would be more appropriate.”

- Rick B.

7 Mar 01Jan 20057 Mar 01Jan 2005


Abstract or SpecificAbstract or Specific

Which style are you most likely to use in an LDM / PDM?1. The more flexible (Abstract) the better2. 3. 4. 5. The more precise (Specific) the better

Which style are you most likely to use in an LDM / PDM?1. The more flexible (Abstract) the better2. 3. 4. 5. The more precise (Specific) the better

7 Mar 01Jan 20057 Mar 01Jan 2005


To Party or Not to PartyTo Party or Not to Party

• Party, Party Role, Party Type, Party Category

• Migration of Party ID• Use of Subtype “Owned Keys”• Universality of Party

• Party, Party Role, Party Type, Party Category

• Migration of Party ID• Use of Subtype “Owned Keys”• Universality of Party

7 Mar 01Jan 20057 Mar 01Jan 2005



• Party

• Non-Party

• Party

• Non-Party

PARTY ROLE

PARTY

VENDOR

CUSTOMER

DISTRIBUTOR

EMPLOYEE

VENDORCUSTOMER

ORGANIZATION

PERSON

7 Mar 01Jan 20057 Mar 01Jan 2005



It is so unfortunate that there is still a mentality that we need to model the simple case only. I personally think that once you understand the party, party role model it is simpler. That simplicity reinforces itself as you are able to apply the same pattern to lots of different situations and environments.

- George P.

It is so unfortunate that there is still a mentality that we need to model the simple case only. I personally think that once you understand the party, party role model it is simpler. That simplicity reinforces itself as you are able to apply the same pattern to lots of different situations and environments.

- George P.

7 Mar 01Jan 20057 Mar 01Jan 2005



The problem I find with PARTY ROLE is that it is derivable information. This is purely a classification of an individual body based on other data that must be available. Otherwise if I identify you as my “customer” what business meaning does it have? There must be one or more “orders”, or “contracts”, or “correspondences”, or “contacts”, or whatever I deem necessary for it to be sensible for you to be known as my customer. It’s the ORDER that makes you my customer, not being my customer that makes me able to take your ORDER.

- Mike V.

The problem I find with PARTY ROLE is that it is derivable information. This is purely a classification of an individual body based on other data that must be available. Otherwise if I identify you as my “customer” what business meaning does it have? There must be one or more “orders”, or “contracts”, or “correspondences”, or “contacts”, or whatever I deem necessary for it to be sensible for you to be known as my customer. It’s the ORDER that makes you my customer, not being my customer that makes me able to take your ORDER.

- Mike V.

7 Mar 01Jan 20057 Mar 01Jan 2005


To Party or Not to Party - Vote

To Party or Not to Party - Vote

Which is the best approach to PARTY in an LDM /PDM?

1. It’s the only way to be fully logical 2. 3. 4. 5. It’s too academic, too theoretical and it

just won’t work

Which is the best approach to PARTY in an LDM /PDM?

1. It’s the only way to be fully logical 2. 3. 4. 5. It’s too academic, too theoretical and it

just won’t work

7 Mar 01Jan 20057 Mar 01Jan 2005


DA – Lawyer or Architect?DA – Lawyer or Architect?

• Several new legislative actions focus on data quality, integrity, reliability, availability…

• Sounds like good DA work, right?

• Several new legislative actions focus on data quality, integrity, reliability, availability…

• Sounds like good DA work, right?

7 Mar 01Jan 20057 Mar 01Jan 2005


ExamplesExamples

• Sarbanes-Oxley (Sarbox): Financial Accountability & Reliability

• Anti-Spam Legislation: Data integrity, Reliability

• Privacy: Quality, Reliability, Accountability…

• Sarbanes-Oxley (Sarbox): Financial Accountability & Reliability

• Anti-Spam Legislation: Data integrity, Reliability

• Privacy: Quality, Reliability, Accountability…

7 Mar 01Jan 20057 Mar 01Jan 2005


DA Role in ComplianceDA Role in Compliance

1. Finally, a reason for the Execs to support us…

2. 3. 4. 5. Hey! I don’t even have time

write good definitions…

1. Finally, a reason for the Execs to support us…

2. 3. 4. 5. Hey! I don’t even have time

write good definitions…

7 Mar 01Jan 20057 Mar 01Jan 2005


Did UML kill the Data Modelling Star?

Did UML kill the Data Modelling Star?

• Combines data and process into one diagramming technique

• Some tools support ERDs as well as UML models

• Many DA’s are being pressured to choose between UML and ERDs.

• Combines data and process into one diagramming technique

• Some tools support ERDs as well as UML models

• Many DA’s are being pressured to choose between UML and ERDs.

7 Mar 01Jan 20057 Mar 01Jan 2005


UML Class DiagramUML Class Diagram

7 Mar 01Jan 20057 Mar 01Jan 2005


UML - Vote?UML - Vote?

Is UML an acceptable replacement for LDM?

1. Never – 2 different things2. 3. 4. 5. UML is the future, we might as well

switch now….

Is UML an acceptable replacement for LDM?

1. Never – 2 different things2. 3. 4. 5. UML is the future, we might as well

switch now….

7 Mar 01Jan 20057 Mar 01Jan 2005


Generic Data ModelsGeneric Data Models

• Generic Data Models are models or subsets of models that an organization can purchase.

• They are sometimes prepared for a specific industry

• May be very generic, may be both.

• Generic Data Models are models or subsets of models that an organization can purchase.

• They are sometimes prepared for a specific industry

• May be very generic, may be both.

7 Mar 01Jan 20057 Mar 01Jan 2005


Generic Data Models - Vote

Generic Data Models - Vote

What do you think about Generic Data Models or Patterns?

1. I love them and won’t ever start a project from a blank page again.

2. 3. 4. 5. They are so high level that I don’t see the

value in purchasing them.

What do you think about Generic Data Models or Patterns?

1. I love them and won’t ever start a project from a blank page again.

2. 3. 4. 5. They are so high level that I don’t see the

value in purchasing them.

7 Mar 01Jan 20057 Mar 01Jan 2005


Who Gets to Update the Model?

Who Gets to Update the Model?

• Many marketing pitches portray a team of developers, modelers, DBAs, etc., happily working on the model

• Is it “More hands make for less work” or “Too many cooks spoil the soup”?

• Many marketing pitches portray a team of developers, modelers, DBAs, etc., happily working on the model

• Is it “More hands make for less work” or “Too many cooks spoil the soup”?

7 Mar 01Jan 20057 Mar 01Jan 2005


Who Gets to Update the Model

Who Gets to Update the Model

1. Only seasoned Data Modeling Professionals

2. 3. 4. 5. Everybody – we’re all

professionals here

1. Only seasoned Data Modeling Professionals

2. 3. 4. 5. Everybody – we’re all

professionals here

7 Mar 01Jan 20057 Mar 01Jan 2005


Who calls the shots in a Logical Data Model?


• Many stakeholders in a LDM• Not everyone shares the same

understanding about the purpose of the LDM

• DM tools can greatly influence the decision

• DA’s can report through a variety of departments

• Many stakeholders in a LDM• Not everyone shares the same

understanding about the purpose of the LDM

• DM tools can greatly influence the decision

• DA’s can report through a variety of departments

7 Mar 01Jan 20057 Mar 01Jan 2005




The FINAL say on LDM / PDM decisions belongs to:1. The Data Architect/Modeller2. The Project Manager3. The User…Customer4. The DBA5. The CIO/CEO

The FINAL say on LDM / PDM decisions belongs to:1. The Data Architect/Modeller2. The Project Manager3. The User…Customer4. The DBA5. The CIO/CEO

7 Mar 01Jan 20057 Mar 01Jan 2005


XML – the next Data Specification


• XML is a type of markup language for data

• Similar to HTML, but carries meaning as well as format metadata

• Many who use EDI are looking toward XML

• Many vendors are jumping onto this ‘standard’

• Still requires establishing meaning.

• XML is a type of markup language for data

• Similar to HTML, but carries meaning as well as format metadata

• Many who use EDI are looking toward XML

• Many vendors are jumping onto this ‘standard’

• Still requires establishing meaning.

7 Mar 01Jan 20057 Mar 01Jan 2005


XMLXML

<?xml version="1.0" encoding="iso-8859-1" ?> - <clientapp> <global>

<string phone_number="*8369464" /> <string user id=“TestUser”/><string dir_html="html" /> <string dir_language="english" /> <string filename_html_connected="connected.htm"

/> <string filename_html_targetfile="target.htm" />

</global>-

<?xml version="1.0" encoding="iso-8859-1" ?> - <clientapp> <global>

<string phone_number="*8369464" /> <string user id=“TestUser”/><string dir_html="html" /> <string dir_language="english" /> <string filename_html_connected="connected.htm"

/> <string filename_html_targetfile="target.htm" />

</global>-

7 Mar 01Jan 20057 Mar 01Jan 2005




I have heard from a wide variety of sources that XML means that data modelling is dead -- who needs a data model when you can create a DTD and call the piece of data anything you want? All you have to do is convince everyone else to call it the same thing...

- Karen Lopez

I have heard from a wide variety of sources that XML means that data modelling is dead -- who needs a data model when you can create a DTD and call the piece of data anything you want? All you have to do is convince everyone else to call it the same thing...

- Karen Lopez

7 Mar 01Jan 20057 Mar 01Jan 2005




XML is very simple. It is a facility for defining your own tags in transmitting documents. (With HTML you have to use theirs.) The implication of this is that the browser/person receiving the document has to know how to interpret your tags. This is the fun part. Some organizations have begun to define a set of tags to serve their purposes -- mathematicians, chemists, metadata gurus, etc.

Alternatively, if you send the "Document Tag Definition (DTD)" with your data, it can be used to decode it, at least syntactically.

- David H.

XML is very simple. It is a facility for defining your own tags in transmitting documents. (With HTML you have to use theirs.) The implication of this is that the browser/person receiving the document has to know how to interpret your tags. This is the fun part. Some organizations have begun to define a set of tags to serve their purposes -- mathematicians, chemists, metadata gurus, etc.

Alternatively, if you send the "Document Tag Definition (DTD)" with your data, it can be used to decode it, at least syntactically.

- David H.

7 Mar 01Jan 20057 Mar 01Jan 2005




I see XML as a new medium for presenting data. To effectively present the data it needs to be understood and this is the role of the data model. Do we have the fundamental meaning of our data understood and represented in a quality data model? If yes then the transformation of the requirements into the presentation by XML will be much more effective. ….Enough said for now on how I feel that the data model and XML need to correlate data requirements.

- Paul E.

I see XML as a new medium for presenting data. To effectively present the data it needs to be understood and this is the role of the data model. Do we have the fundamental meaning of our data understood and represented in a quality data model? If yes then the transformation of the requirements into the presentation by XML will be much more effective. ….Enough said for now on how I feel that the data model and XML need to correlate data requirements.

- Paul E.

7 Mar 01Jan 20057 Mar 01Jan 2005


XML - Vote?XML - Vote?

Will XML replace data models?1. Never – 2 different things2. 3. 4. 5. XML is the future, we might as

well switch now….

Will XML replace data models?1. Never – 2 different things2. 3. 4. 5. XML is the future, we might as

well switch now….

7 Mar 01Jan 20057 Mar 01Jan 2005


Is that with one “L” or Two?

Is that with one “L” or Two?

How is it spelled?1. Modeling/Modeler2. Modelling/Modeller3. It depends on my mood…

How is it spelled?1. Modeling/Modeler2. Modelling/Modeller3. It depends on my mood…

7 Mar 01Jan 20057 Mar 01Jan 2005


Online Data Modelling Resources

Online Data Modelling Resources

• InfoAdvisors (http:///www.infoadvisors.com)

– The Data Model List– Product User Discussion Groups (ERwin,

Advantage Repository, Visible Analyst/Advantage, ER/Studio, DBArtisan, CaseWise)

• The Data Administration Newsletter (http://www.tdan.com)

• InfoAdvisors (http:///www.infoadvisors.com)

– The Data Model List– Product User Discussion Groups (ERwin,

Advantage Repository, Visible Analyst/Advantage, ER/Studio, DBArtisan, CaseWise)

• The Data Administration Newsletter (http://www.tdan.com)



7 Mar 01Jan 20057 Mar 01Jan 2005


Recommended BooksRecommended Books

• Data Modeling Essentials, Graeme Simsion, 2004

• Building Quality Databases with IDEF1X, Thomas Bruce, 1992, Dorset House, ISBN 0-932633-18-8

• The Data Modeling Handbook, Michael

Reingruber & William Gregory, 1994, Wiley QED, ISBN0-471-05290-6

• A Practical Guide to Logical Data Modeling, George Tillmann, 1993,

McGraw-Hill, $45, ISBN 0-07-064615-5

• Data Modeling Essentials, Graeme Simsion, 2004

• Building Quality Databases with IDEF1X, Thomas Bruce, 1992, Dorset House, ISBN 0-932633-18-8

• The Data Modeling Handbook, Michael

Reingruber & William Gregory, 1994, Wiley QED, ISBN0-471-05290-6

• A Practical Guide to Logical Data Modeling, George Tillmann, 1993,

McGraw-Hill, $45, ISBN 0-07-064615-5

7 Mar 01Jan 20057 Mar 01Jan 2005


Recommended BooksRecommended Books

• Data Model Patterns: Conventions of Thought, David C. Hay, 1996, Dorset House, ISBN 0-9326333-29-3

• The Data Model Resource Book, Silverston, Volumes 1 and 2

• Data Model Patterns: Conventions of Thought, David C. Hay, 1996, Dorset House, ISBN 0-9326333-29-3

• The Data Model Resource Book, Silverston, Volumes 1 and 2

Jan 2005Jan 2005Jan 2005Jan 2005


InfoAdvisorsInfoAdvisors11066 Sheppard Ave EastToronto, ON CANADA

[email protected]://www.infoadvisors.com

8000+ IRM participants subscribed to several discussion groups

11066 Sheppard Ave EastToronto, ON CANADA

[email protected]://www.infoadvisors.com

8000+ IRM participants subscribed to several discussion groups

mailto:[email protected]


mailto:[email protected]


Documents

Data Modelling Contentious Issues The Best Debates from the Data Modelling List January 2005 Karen Lopez InfoAdvisors.com The Best Debates from the Data