Access Layer Overview v6a - Module 5

8/20/2019 Access Layer Overview v6a - Module 5

1/12

1 3/7/16 Teradata Confidential

1. How well do the characteristics of the model support the typeof model?

2. How well does the model capture the requirements?

3. How complete is the model?

4. How structurally sound is the model?

5. How well does the model leverage generic structures?

6. How well does the model follow naming standards?

. How well has the model !een arranged for reada!ility?

". How good are the definitions?

#. How consistent is the model with the enterprise?

1$. How well does the metadata match the data?

Data Model Scorecard®: Semantic Data

Model Validation


2/12


• Model type category>Ensures that the definition for the model type is

met

• Semantic data model red flags>Normalized structure for analytics>There are subtypes on the model>Fuzzy grain

! "o# $ell Do the%haracteristics of the ModelSupport the Type of Model&

'elational Dimensional%onceptual

(ogical

)hysical

?


3/12


• %orrectness category>Ensures the model accurately represents the

business and supports the application re*uirements

• Semantic data model red flags>Data (ab not le+eraged

>,seless aggregates

>Data elements #ith a data modeling tool defaultformat

-! "o# $ell Does the Model%apture the 'e*uirements&


4/12


• %ompleteness category>Ensures content and metadata are not leading

to#ards scope creep or .)hase //0

• Semantic data model red flags>Missing precise and complete set of business

*uestions

>

Missing definitions>Missing nullability

1! "o# %omplete is the Model&


5/12


• Structure category>Ensures that the model follo#s good design

principles• Semantic data model red flags>%ircular relationships

>)artial 2ey relationship

>Data element appears more than once insame model3 but #ith different format orlength!

4! "o# Structurally Sound is theModel&


6/12


5! "o# $ell Does the Model(e+erage 6eneric Structures&

• 7bstraction category>Ensures the correct le+el of generic concepts

are applied on the model

• Semantic data model red flag>Too abstract8

THING


7/12


9! "o# $ell Does the ModelFollo# Naming Standards&

• %onfirms that structure3 term3 andsynta ha+e been correctly applied tothe model

CUST_LAST_NAM

Structure

One subjectarea, zero or

moremodifiers, one

class word

TermProper

language andabbreviations

%ynta&'lural or singular( whether hyphens(

spaces( or )amel!ac*( and case


8/12


;! "o# $ell "as the Model Ensures the model is easy to read8

•Semantic data model red flags>Dimensions do not surround the fact table

>7 font that is too small

>Too many entities on one display


9/12


• %onfirms all definitions are clear3complete3 and correct!

?! "o# 6ood are theDefinitions&

Formal

54

+he )ustomer ,dentifieris the identifier for the

customer

An employee is someone wo

breates o!ygen


10/12


•%onsistency category>Ensures the model complements the .SMDimension not conformed

>

/mpossible to source from integrated data layer

@! "o# %onsistent is the Model$ith the Enterprise&


11/12


•Data category>Ensures structure matches reality

•Semantic data model red flags>Data does not match definition3 name3 format3 or

length of metadata!

>Data sparse in re*uired data element

>Data duplicates in a uni*ue data element

A! "o# $ell Does the MetadataMatch the Data&

CALENDAR

Month CodeYear Code

Month NameMonth Sequence Number A!"#"$Year NameYear Sequence Number

%R&DUCT

%roduct Ident'('er

%roduct Name A!"#"$%roduct )ro*en Ind'cator A!"#+,IE"#"$%roduct U%C Code A!+#"$%roduct EAN Number A!-#"$%roduct L'ne Name%roduct Cate.or/ De0cr'1t'on Te2t

GE&GRA%HY

Re.'on Ident'('er

Re.'on Name A!"#"$Countr/ Ident'('er Countr/ NameCountr/ IS& Code

SALES

%roduct Ident'('er )!$Month Code )!$Year Code )!$Re.'on Ident'('er )!$

Gro00 Sa3e0 Amount

-&tract filesata!ase ta!les

ata entry screens


12/12


1. How well do the characteristics of the model support the type

of model?

2. How well does the model capture the requirements?

3. How complete is the model?

4. How structurally sound is the model?

5. How well does the model leverage generic structures?

6. How well does the model follow naming standards?

. How well has the model !een arranged for reada!ility?

". How good are the definitions?

#. How consistent is the model with the enterprise?

1$. How well does the metadata match the data?

Data Model Scorecard®: Semantic DataModel Validation

Access Layer Overview v6a - Module 5

Documents