the universal database design the future of data for database architects
TRANSCRIPT
The Universal Database Design
The Future of Data
For Database Architects
To handle objects effectively, you need to be able to relate any entity to any other entity, but this is difficult when they are separated into different entity tables.
Primary Entity Key
Relationship Type
Secondary Entity
Key
(Fan23) Type Of (AcmeFan)
(Fan23) Contains (Motor23)
(Fan23) Connected To
(Wire 14)
Fan Table
Motor Table
Wire Table
Primary /
Secondary
FansMotor
sWires
Fans X X X
Motors
X X X
Wires X X X
This chart shows that to relate the entities of an N=3 number of tables to each other, there are N2=9 number of combinations – a measure of complexity.
EntityKey
NameManu
fLengt
h
Length
UnitSize
Size Unit
XX
Unit
(AcmeFan)
Acme Fan
Acme
(Fan23) Fan23 Acme 13 in Small
(Motor23)
Motor23 Nadir 15 kW 30 %
(Wire14)
Wire14 Nadir 10 ft 20 mm2 0.001 Ω
Because database tables are custom-designed and crafted to match the applications they support, they are isolated from each other.
Accounting
Accounting Data
Analysis
Analysis Data
Facilities
Facility Data
Purchasing
PurchasingData
Facility DataFacility
Data
PurchasingData
PurchasingData
Analysis Data
Accounting Data
Accounting Data
Analysis Data
Accounting Data
PurchasingData
PurchasingData
PurchasingData
Facility Data
Facility Data
PLM Data
CAD Data
Analysis Data
Facilities Data
Planning Data
Purchasing Data
Analysis
PLM
CAD
Facilities
Planning
Purchasing
Integration leads to an N2 number of point-to-point translations and increased developmental gridlock.
Because most changes are structural, it increases the cost of change, including the transition (or loss) of legacy data.
The General Relation Table uses foreign keys from a General Entity Table holding the entity and relationship types.
Primary Entity Key
Relationship Type
Key
Secondary Entity
Key
(Fan23) (TypeOf) (AcmeFan)
(Fan23) (Contains) (Motor23)
(Fan23) (ConnectedTo)
(Wire 14)
To handle object data and other types, we need to increase the flexibility of relationships, but without increasing the complexity.
As shown previously, complexity of a system is proportional to the number of different types of tables squared(N2). To reduce the complexity, it is very effective to reduce the number of table types.
A solution starts with defining a General Relation Table to handle any standard relationship between two entities.
Entity Key Name ManufLengt
h
Length
UnitSize
Size Unit
XX
Unit
(AcmeFan) Acme Fan Acme
(Fan23) Fan23 Acme 13 in Small
(Motor23) Motor23 Nadir 15 kW 30 %
(Wire14) Wire14 Nadir 10 ft 20 mm2 0.001 Ω
(TypeOf) Type Of
(Contains) Contains
(ConnectedTo)
Connected To
A General Relation Table requires entities and relationship types to be from the same table, the General Entity Table.
Combining all entities in the same table lets any entity have any attribute, but generates more unused space as is.
Using normalization, attribute columns can be replaced with relationships, eliminating the empty space, and allowing multiple values for each attribute.
Primary Entity Key
Relationship Type
Key
Secondary Entity
Key
(Fan23) (TypeOf) (AcmeFan)
(Fan23) (Contains) (Motor23)
(Fan23) (ConnectedTo)
(Wire14)
(AcmeFan)
(Manuf) (Acme)
(Fan23) (Manuf) (Acme)
(Motor23) (Manuf) (Nadir)
(Wire14) (Manuf) (Nadir)
Entity Key Name
(AcmeFan) Acme Fan
(Fan23) Fan23
(Motor23) Motor23
(Wire14) Wire14
(TypeOf) Type Of
(Contains) Contains
(ConnectedTo)
Connected To
(Manuf) Manuf
Count
42Part123
PartNo
Parts Catalog Inventory Table The old design pattern holds values
whose meaning are dependent on the row and column positions in each table. This leads to separate, hard-coded, custom-built functions to handle each different table and column.
The new design pattern holds small, atomic pieces of Equal Format Data in a standard container, called an Equal Format Database. This leads to standard, reusable functions with powerful data management capabilities across all “columns”.
Part123 Count 42
Equal Format Data Atom
Equal Format
Database
To see the differences in normalization patterns, first examine the statistics of the entities and attributes used.
Fans
Motors
Wires
The old design method uses the pattern of data to define separate tables that match the groupings, each with some of the possible attributes.
The new design method keeps everything in the same “table”, with a full selection of all possible attributes – past, present and future.
Fan Table
Motor Table
Wire Table
Old
New
Entity-Relationship (E-R) Triplets have some limitations. They can be difficult to use for some data. It would help to expand past triplets, but we need a larger pattern to guide us.
Subject Key
Verb Key Object Key
(Fan23) (TypeOf) (AcmeFan)
(Fan23) (Contains) (Motor23)
(Fan23) (ConnectedTo)
(Wire14)
(AcmeFan) (Manuf) (Acme)
(Fan23) (Manuf) (Acme)
(Motor23) (Manuf) (Nadir)
(Wire14) (Manuf) (Nadir)
The first step in improving triplets is to convert the entity-relationship triplets to Semantic Triplets. This only requires a name change to the columns, making each row into a Sentence. This concept lays the foundation of the next steps.
Sentence Key
Syntax Key Phrase Key
(S1) (Subject) (Fan23)
(S1) (Verb) (TypeOf)
(S1) (Object) (AcmeFan)
(S2) (Subject) (Fan23)
(S2) (Verb) (Contains)
(S2) (Object) (Motor23)
(S3) (Subject) (Fan23)
(S3) (Verb) (ConnectedTo)
(S3) (Object) (Wire14)
(S4) (Subject) (AcmeFan)
(S4) (Verb) (Manuf)
(S4) (Object) (Acme)
Another normalization step moves each phrase to its own row, allowing more phrases per sentence and more phrases per syntax position.
Now you can scan through “tables” and “columns” as easily as rows.
Controls can be added for special capabilities, such as nondestructive changes, point-in-time storage, with less locking problems.
Equal Format
Database
Analysis
PLM
Facilities
CAD Planning
Purchasing
PLM Data
CAD Data
Analysis Data
Facilities Data
Planning Data
Purchasing Data
Analysis
PLM
CAD
Facilities
Planning
Purchasing
The old design pattern leads to an N2 number of point-to-point translations, leading to developmental gridlock.
The new design pattern reduces the burden of point-to-point translations, providing a uniform container, a uniform pipeline, and simpler systems.
EFD
EFD
EFD
EFD
EFD
EFD
EFD
Analysis
PLM
Facilities
CAD Planning
Purchasing
EFDEFD
Equal Format Database containers can be grouped together to take advantage of many scaling techniques, without altering the data atoms moved between them.
Built on standard database systems, Equal Format Databases are easily integrated with legacy data in various ways.
ProjectSystem
SecuritySystem
DataSystem
VendorSystem
CustomerSystem
Legacy Database
Legacy Database