Logical Data Modelling

The sole purpose of an Information System is to support or automate business activities by storing and processing relevant business information or data. It is therefore critical to the success of any IS development that the meaning, structure and business rules of the required data are fully analysed, understood and modelled. During the Investigation phase we are concerned with understanding the underlying (i.e. logical) data requirement rather than making decisions about its physical implementation.

Current and Required Data

Most of the data required for the future system will be the same in content or meaning as that used currently. The other form of data analysis concerns requirements for new business data. The approach of starting with an analysis of existing data (or even re-using existing data models) will provide the most rigorous and efficient approach. This approach will also help in driving out restrictions in data support arising from existing technical constraints.

Physical vs Logical Data Structures

An organisations data will be physically stored in many different places, e.g. paper files, computer files. This data will almost inevitably contain duplications and compromises due to the physical restrictions of storage, processing or practicality. Example:
A physical purchase order form will hold information about products (product name, product number, product price), suppliers (supplier name, supplier address), the orders heading (purchase order number, purchase order date) as well as the quantity of each product ordered (quantity ordered). While we may have a single physical grouping of data on one purchase order form, what we actually have is information about several different things - products, suppliers, and purchase orders. In other words the underlying logical view is of a number of separate data groupings, each describing a different business concept or object. We will also find that information on, for example, products is physically held in many other places, such as on customer orders, invoices and despatch notes. This all leads to a confusing mess of duplication and interconnecting information, which in turn leads to problems in maintaining data consistency and integrity.

The LDM
In SSADM the vehicle for analysing the logical structure of an organisations information is the Logical Data Model (LDM). A Logical Data Model is a way of graphically representing what that information is really all about, how it relates to other information and business concepts, and how business rules are applied to its use in the system. The LDM is possibly the most important and ultimately the most rigorous product of an entire SSADM project. Logical Data Models consist of two parts:
a diagram called the Logical Data Structure (LDS); a set of associated textual descriptions that explain each part of the diagram.

Entities
Any object or concept about which a system needs to hold information is known as an Entity Type (or entity for short). To be a valid entity we must wish to hold information on more than one occurrence of it. Entity occurrences are real world instances of an entity type. For example the entity type Supplier will have occurrences such as:
Supplier No Supplier Name Supplier Address 3621 Off Beat Recordings 12 High Street etc. 2327 Bella Sonic Lake Industrial Estate etc.

Entities (continued)
The symbol for an entity in an LDS is a round cornered rectangle containing the entitys name (which must be unique):

Supplier unique name

An entity must have a number of properties to qualify as such: - There must be more than one occurrence of the entity. - Each occurrence should be uniquely identifiable. - There must be data that we want to hold about the entity. - It should be of direct interest to the system.

Attributes
Each item of information (or data) that we hold about an entity is known as an attribute or data item. Examples of attributes for Supplier might be supplier number, supplier name, supplier address, and supplier telephone no. The detail of an entitys attributes is not formally included on the LDS itself. This is held in separate textual descriptions, which will be discussed later.

Relationships
Entities do not exist in isolation, but are related to other entities. In physical data structures these relationships are signified by physical links such as pointers or placement in the same file or document. In logical models relationships represent business associations or rules and not physical links. Any entities that are related are linked by a line on the LDS. The line is labelled with the name of the relationship, and is named in both directions.
supplier for Supplier ? placed with ?

Purchase Order

Degree
The number of occurrences of each entity type participating in a given relationship is denoted by the degree or cardinality of that relationship, and illustrated on the LDS by adding crows feet to the relationships line.
A

m:n

1:m

1:1

There are three types of degree:

Many to Many (m:n). This tells us that each occurrence of A is related to one or more occurrences of B, and each occurrence of B is related to one or more occurrences of A. One to Many (1:m). This tells us that each occurrence of A is related to one or more occurrences of B, but each occurrence of B is related to only one occurrence of A. One to One (1:1). This tells us that each occurrence of A is related to only one occurrence of B, and each occurrence of B is related to only one occurrence of A.

Optionality
Each relationship is further annotated to show if it must exist for all occurrences of the participating entity types. If there can be occurrences of one entity that are not related to at least one occurrence of the other, then the relationship is said to be optional for that entity. The relationship line is then converted to a dashed line at its optional end (which could mean both ends if both entities are optional participants).
Supplier

supplier for

placed with

Purchase Order

Developing the LDS

To start with we are only interested in producing a high level model of the current systems underlying data structure. Due to its largely conceptual nature Logical Data Modelling can be one of the most intense activities of an SSADM project. In many projects development of the LDM is started by holding brainstorming sessions with small groups of analysts and users. With a little practice analysts often find that the best method of data modelling is to draw up possible LDSs almost instinctively Relationships are added as each entity is identified and then checked with users on the spot. This approach has a lot to recommend it, particularly at this level of detail or for small systems, as diagrams are produced and verified quickly.

Identifying Entities
To identify entities in the current environment we can begin by looking at our physical data stores to find out exactly what it is that they hold information about. If we take the customer order file and discuss it with users, we find that it not only contains details of each individual order, but of the customers themselves,
i.e. customer address, customer telephone number etc., and so encompasses at least two entities, namely Customer and Customer Order.

Supplier Product

Purchase Order Stock

Customer Customer Order

Delivery (from a supplier) Despatch (to customers)

Verification
Once the list has been drawn up we should verify it with key users during preliminary scoping interviews. The key questions to ask of each entity are:
Are any of the candidates merely attributes of another entity? Do any of the candidates represent a subset of occurrences of another entity? Do all of the entities have a unique identifier?

During this process we may discover new entities, merge existing entities or discard candidates as being outside the area of investigation.

Note: There will often be relationships between entities that exist in the real world, but which are not of relevance to the system under discussion. E.g a customer of ZigZag may well be employed by one of its suppliers. This is NOT something that ZigZag will be interested in recording!

Adding Relationships
We now examine each entity to see if it is directly related, in a way that is of interest to the system, to any of the other entities. The best way to do this is in discussion with users, either taking each entity in turn, or starting with a key entity and moving around the LDS network as the relationships are identified. Having identified where we think relationships exist, we now consider their degree, optionality and names. We do this by identifying the business rules that apply to each entity pairing. The basic process is the same for all pairings, so we will look at just one example.

Stock - Delivery
We first consider the relationship from the Stock perspective:
Each Stock occurrence will consist of a quantity of a single product, all of which was delivered on the same delivery. If within the depot we have a quantity of a given product, some of which was delivered in one delivery and some in another, then we will have more than one Stock. This is an example of one of ZigZags business rules, and one that will continue in the new system.

Delivery

Thus each Stock occurrence is related to just one Delivery.

Each delivery may contain a number of different products, each of which will be stored as a separate stock (remember that each Stock occurrence is a quantity of a single product).

Stock

Thus each Delivery is related to one or more Stock occurrences.

Stock Delivery (continued)

We now consider the optionality of the relationship:
Each Stock must have been delivered by a Delivery. So the relationship at the Stock end is mandatory. However a Delivery could be rejected for quality reasons by the depot, in which case the delivery would be recorded but would not be related to any subsequent Stock occurrences. So the relationship is optional at the Delivery end.
Delivery

Stock

Choosing a name is often the hardest part of the procedure. It is important to name a relationship in both directions as it forces us to examine the true nature of the relationship, sometimes leading to the discovery of additional relationships or even entities. We should always try to choose phrases that accurately reflect the users view of the relationship. In our example it is not too difficult to find reasonable names: delivery of and delivered by.

Delivery

delivery of

delivered by

Stock

Overview LDS
Continuing this process for all of the relationships identified on the matrix gives us a first-cut overview LDS for the current system:
supplier of Supplier supplier of

supplier for

Despatch

Customer

Delivered by placed with Purchase Order results in ordered by order for held as holding of delivered by Delivery delivery of Stock supplied by

despatch of despatched in ordered by Product customer order for

orderer of

ordered by

Customer Order

result of

Drilling Down.
The overview LDS provides us with a good basis for building a more complete model of current data. We begin the process of creating a detailed model by looking at this model and discussing it with users to check our understanding of the scope of current data and to uncover lower level entities which can be added immediately.
Product Type

Depot Zone

Product

Supplier Invoice

Delivery

Masters and Details

Most relationships are 1:m. The entity at the 1 end is known as the master and the entity at the m end as the detail.
Master

Detail

Supplier

The terms master and detail refer only to an entitys role in a particular relationship. It is quite possible for an entity to be the master in one relationship and the detail in another.

Product

Stock

Keys
We should be able to select at least one identifier for each entity type,
i.e. an attribute that enables each occurrence of an entity to be uniquely identified, e.g. for Customer we could use customer number.

Any attribute or set of attributes which together uniquely identify an entity is known as a candidate key. One of these candidates (there will often only be one) should be selected as the primary key. Whenever we require direct access to an entity, the primary key is used to identify which occurrence we are interested in. For example, if we needed to access the Supplier entity to find out a suppliers address, we would use the primary key of supplier number to identify the correct occurrence.

Foreign Keys
If we have a relationship between two entities we need to be able to associate the occurrences at one end with the related occurrences at the other. In a relational model (such as the LDM) we do this by including the primary key of the master in the set of attributes of the detail. The copy of the masters primary key in the detail entity is known as a foreign key.

Key Navigation
Supplier attributes:
Supplier No. (Primary Key), e.g 271 Supplier Address etc.
Supplier

Purchase Order attributes:

P.O. Number (Primary Key), e.g 5001 P.O. Date etc. Supplier Number (Foreign Key), e.g 271
supplier for

To access all purchase orders placed with supplier number 271, we look for all occurrences of Purchase Order with a supplier number attribute value of 271. Coming in the opposite direction, to access the supplier for purchase order 5001, we look for the single occurrence of the Supplier entity whose primary key is equal to the supplier number given in the foreign key of purchase order number 5001, i.e. supplier number 271.

placed with Purchase Order

Types and Notation

Primary Keys belong to one of three types:
1. A Simple Key, consisting of a single attribute; 2. A Compound Key, consisting of two or more foreign keys; 3. A Hierarchic or Composite Key, consisting of one or more foreign keys and a qualifying non-foreign key attribute.

Notation
The primary key is underlined and the foreign key preceded by an asterisk to show the contents of each entity: Supplier (supplier number, supplier address, supplier tel. no.) Purchase Order (P.O. number, P.O. date, *supplier number)

Resolving Many-to-Many Relationships

Many design techniques can only be carried out on hierarchical (i.e. master-detail) relationships which are hidden by m:n relationships. m:n relationships make navigation around the model very difficult or even impossible (and, although we are not really concerned with technical issues at this point, they cannot be implemented). m:n relationships very often hide information about the participating entities or the relationships themselves.

Resolving Many-to-Many Relationships - example

Each Product may be ordered by one or more Purchase Orders. Each Purchase Order must be an order for one or more Products. So where do we place the quantity ordered?

Resolving Many-to-Many Relationships - example

If we look at a sample purchase order of ZigZag, we will discover that details of quantities and products are held in individual purchase order lines.

Purchase Order Number:

0021113

Purchase Order Date:

4/3/01

Supplier:
2327 Bella Sonic Lake Industrial Estate Unit 5 NE3 7AJ

Delivery Address:
Depot 1 Harrow Way Harrow HA4 3NB

Qty

Your Product Ref

BJB001 3485VHS/3

Our Product

Description

Format

Unit Price

Ref
100 500 884690 993201 The Best of Johnnie Boy Unbranded Blank 3hr Video Tapes CD BV 6.99 0.53

Resolving Many-to-Many Relationships - example

So in this case we can choose a natural link entity, which we will call Purchase Order Item. Purchase Order Line sounds a bit too similar to the physical printed line on the order form. The key for Purchase Order Item will be Purchase Order Number plus Product Number a compound key.
Purchase Order

Product

contains contained in

ordered by order for

Purchase Order Item

Relationships in M:N Resolutions

Whenever we introduce a link entity we need to ensure that the relationships we recorded previously with its master entities are still valid. For example, in our overview LDS we recorded a many to many relationship between Despatch and Customer Order. This may at a high level appear reasonable as it is common for some items in an order to go into one van load (Despatch) and some into another. However, the contents of each item within the order is always despatched in its entirety in the same van load (i.e. if 3 copies of Puccinis Tosca are ordered within a single customer order, they will all be delivered together). Therefore, each Despatch is actually related to many Customer Order Items, rather than to whole Customer Orders.

Link Entities
Depot Zone and Product Type provide another more complex illustration of many to many relationships: Each Depot Zone may store one or more Product Types. Each Product Type must be storable in one or more Depot Zones. The attributes that make up Depot Zone are Depot Zone Number, Shelf Height, and Depot Zone Description etc. Depot Zone Number is a unique identifier that is assigned to each Depot Zone, and is the label attached to the end of each row of shelving in the zone. Depot Zone Description would include values such as CD and DVD, Videos and Books, and Tape etc, which describe the sorts of products that the shelving in each zone can accommodate. The attributes of Product Type include Product type code and Product type name, where the Product type code is an abbreviation of the Product type name, e.g. BV for Blank Video, DVD for DVD etc. So, for example, we might have the following cases: Depot Zones 101 and 105 store DVD and CD product types; Depot Zone 102 stores VHS, BV and SPB (small paperback book) product types.

Link Entities
To make these associations we would have to set up lists of foreign keys in both entities, of arbitrary length.
Significant maintenance overhead Navigation around the model very difficult Against the rules of relational data modelling

Solution: A link entity

Each occurrence will store a valid association or pairing of a Depot Zone occurrence with a Product Type occurrence, such as: Depot Zone Product Type Product 101 DVD Depot Zone Type 101 CD 105 DVD allocated by stored in 105 CD allocation for allocates 102 VHS storage for 102 BV 102 SPB Depot Zone
Allocation

Pigs Ear
substitute for

Product

substituted by

sub with Product Substitute sub of

sub by Product sub for

Resolving One-to-One Relationships

The problems associated with 1:1 relationships are less clear-cut than with m:n relationships:
1:1 relationships often obscure an underlying single entity. There may be a missing link entity. Later design techniques may require all relationships to be master-detail.

In the ZigZag overview LDS there are two 1:1 relationships - between Delivery and Purchase Order and between Supplier Invoice and Delivery .
Deliveries are identified by the purchase order they are satisfying The only information currently held about them details which parts of the purchase order they have successfully delivered. It is quite easy in this case to view Delivery as a logical extension (or conclusion) of a Purchase Order, so we will merge the two entities and transfer all of Deliverys relationships to Purchase Order. To do this successfully, Purchase Order will contain attributes delivery date and suppliers delivery reference while Purchase Order Item will contain quantity delivered.

Resolving One-to-One Relationships (continued)

Supplier Invoice
Purchase Order

Invoice Item

Purchase Order Item

Supplier Invoice

Purchase Order

Purchase Order Item

Resolving One-to-One Relationships (continued)

Supplier Invoice Purchase Order

Purchase Order Item

Purchase Order

Supplier Invoice

Purchase Order Item

Removing Redundant Relationships

One of our aims when drawing up an LDS should be to include only the minimum number of relationships needed to apply all of the business rules relating to data. Any unnecessary relationships are termed redundant, and will involve us in a maintenance overhead if implemented. The major difference between relationships and a route map is that each relationship carries with it a meaning, and so different routes between entities will often have different meanings, or enforce different rules.

Customer

Customer Order

Is this relationship Redundant?

Customer Order Item

Removing Redundant Relationships

Each Purchase Order may be related to a number of Supplier Invoice, each of which is related to a PO Item. Each PO Item may relate to just one Supplier Invoice, which relates to just one Purchase Order. HOWEVER Each Purchase Order MUST contain at least one PO Item. If the Invoice is not present then removing the direct relationship would mean that a relationship could not be established between Purchase Order and PO Item.

Purchase Order

Is this relationship Redundant?

Supplier Invoice

Purchase Order Item

Supplier

Product Type

Depot Zone

Depot Zone Allocation

Product Purchase Order Despatch

Product Substitute

Customer

Supplier Invoice Customer Order

Purchase Order Item Customer Order Item

Stock

Selected ZigZag Entities and Attributes

PRODUCT Product Number *Product Type Code Product Name Product Description Release Date Sell-by Date (special promotional products) Sell-from Date Standard Purchase Price Standard Selling Price PRODUCT SUBSTITUTE *Product Number [substitute] *Product Number [substituted] PRODUCT TYPE Product Type Code Product Type Name Product Type Description DEPOT ZONE Depot Zone Number Depot Zone Description PURCHASE ORDER Purchase Order Number

*Supplier Number
Suppliers Delivery Reference Purchase Order Date Purchase Order Status Delivery Date Delivery Start Time Delivery End Time

SUPPLIER Supplier Number Supplier Name Supplier Address Supplier Tel. No. Supplier Contact Name SUPPLIER INVOICE *Purchase Order Number Suppliers Invoice Number Invoice Date

PURCHASE ORDER ITEM *Purchase Order Number *Product Number *Invoice Number Quantity Required Quantity Confirmed Quantity Delivered Quantity Accepted Invoiced Quantity Agreed Unit Price Required-By Date Required-By Time-Period

STOCK Stock Id *Purchase Order Number *Product Number *Zone Code Quantity Stocked Quantity Stocked Quantity Reserved DEPOT ZONE ALLOCATION *Depot Zone Number *Product Type Code

Completing the Documentation

Entity Description
Entity Name
Purchase Order

Description
Attribute

A request for purchase and delivery of goods from a single supplier.

Primary Key
Yes

Foreign Key

Mandatory/ Optional
M M

Purchase Order Number Purchase Order Date Supplier Number Purchase Order Status Delivery Date Delivery Start Time Delivery End Time

Yes

M M O O O

must/may be
must be may must

either /or

Link Phrase
placed with result in contain

one & only one /one or more

one & only one one or more one or more Min. 6000

Entity Name
Supplier Supplier Invoice P.O. Item

Entity Volumes: User

P.O. Clerk

Max.

15000

Average

10000

Access
Read, Create, Delete, Modify Read Read, Create

Despatch Scheduler Purchaser

Growth Rate: 15% per year

Purchase Orders should be archived to tape six months after the last related line has been delivered or cancelled.

Archiving

Small Projects
Entity Name Supplier Product Depot Zone Purchase Order Purchase Order Item Short Description or Comments (optional) Numbers rise at Xmas Storage Area or Aisle within Depot Numbers rise at Xmas Min Volume 400 10000 26 1500 90000 Max Volume 750 100000 40 16000 420000 Ave Volume 500 25000 32 10000 200000 Growth Rate 5% 25% 20% 15% 25%

Attribute Name Purchase Order Number Purchase Order Date Purchase Order Status

Short Description or Comments (optional) Automatically generated by system Date order placed

Domain Integer DDMMYYYY P(Provisional) V(Placed) C(Confirmed) D(Delivered)

Length 9 8 1

Entity Description Table and Data Catalogue Table for Small Projects

Validating the LDM

we need to check that the LDM can provide access to all of the data items required by each update or enquiry process. Most processes will need to access a number of data items, which will be specified by some selection criteria. These items will often be represented by the attributes of more than one entity. navigate around the relationships of the LDS, applying the selection criteria to filter out the entity occurrences we need to provide all of the necessary data. These navigations are called Access Paths.

Validating the LDM

For example, when allocating a zone in which to store the stock of a particular product received in a delivery (a process called Allocate Stock Zone), we will need to find out which depot zones have been designated for the storage of that type of product. The entry point to the LDS is via the product number in the entity Product. We can then access its product type, and then the possible zones in which this product can be stored by reading through all the occurrences of Depot Zone Allocation for that Product Type.

Product Type

Product

Depot Zone Allocation

Supplier

Product Type

Depot Zone

Depot Zone Allocation

Product

Product Substitute

Purchase Order

Despatch

Customer

Supplier Invoice

Customer Order

Purchase Order Item Customer Order Item

Stock