I have an either/or type of situation in a many-to-many relationship I'm trying to model.
So I have these tables:
Message
----
*MessageID
MessageText
Employee
----
*EmployeeID
EmployeeName
Team
----
*TeamID
TeamName
MessageTarget
----
MessageID
EmployeeID (nullable)
TeamID (nullable)
So, a Message can have either a list of Employees, or a list of Teams as a MessageTarget. Is the MessageTarget table I have above the best way to implement this relationship? What constraints can I place on the MessageTarget effectively? How should I create a primary key on MessageTarget table?
Database in question is SQL Server 2008
So you want to ensure that MessageTargets for a single message all have the employeeID set or the teamID, but not a mixture of both?
Depending on your RDBMS you might be able to create Materialized view and put a constraint on that. The view would look like
select messageId, count(employeeId), count(teamId) from messageTarget
On that you would place a check constraint ensuring that one of the counts is zero.
Alternatively you could replace the MessageTarget with two tables: EmployeeMessageTarget and TeamMessageTarget, each only containing a TargetId and either an EmployeeId in the first table, and a TeamId in the second table.
Your Message table would get two new fields: an EmployeeMessageTargetId and a TeamMessageTargetId plus a check constraint ensuring at least one of those is null. If you make both fields unique you can have a foreign key from the *MessageTarget tables.
The way you present this, it seems that an Employee is-a MessageTarget and also the Team is-a MessageTarget.
So a Message has a Target which is either an Employee or a Team.
Seems to me this is like inheritence (or composition) problem in SQL.
Check this out "Implementing Table Inheritance in SQL Server" .
An Employee is not a MessageTarget per se but perhaps the readings on this can help you on your modelling
Since employees and teams cannot mix in the same message, you'll need to do something like this:
MessageEmployee.MessageIdForEmployee references Message.MessageIdForEmployee.
MessageTeam.MessageIdForTeam references Message.MessageIdForTeam.
And there is the following constraint on the Message table:
CHECK (
(MessageIdForEmployee = MessageId AND MessageIdForTeam IS NULL)
OR
(MessageIdForEmployee IS NULL AND MessageIdForTeam = MessageId)
)
Note how we have a separate junction table for each kind of child table, and junction tables don't reference parent's PK. Instead each junction table references a separate UNIQUE field. Since only one of these fields can be non-NULL, only one kind of child items can be connected to any given message.
NOTE: It is not strictly necessary to match MessageId with MessageIdForEmployee or MessageIdForTeam, but it may simplify querying somewhat.
you may also consider
MessageTarget
----
MessageID
targetID (not nullable)
targetType
then set the type to whichever it should be...
is-a relationships are often instances of the gen-spec pattern. Class Table Inheritance is one way to design tables for cases of gen-spec.
http://martinfowler.com/eaaCatalog/classTableInheritance.html
Consider removing the the "Team ID" field in the MessageTarget entity so that you only have messageId and employeeId. To cater for the team thing (if it is a requirement that each employee must be a member of a team), you can have another entity "Team Messages" where a database trigger will ensure that on insert into that table, you can insert a row in the message target table for each employee in the team. This way you can conveniently link back from each message to an employee from the MessageTarget table or back to a team from the "Team Messages" table. Also gonna make for convenient access in an ORM framework where the Employee entity simply has a List and the Team also has the same
Related
Just curious if I can have the same primary key in 3 different tables? I am going to create an Employee, FullTime and PartTime tables. I would like to make an EmployeeID the primary key for all 3. Any thoughts?
You can have the primary key EmployeeId in a table called Employees. This would have common information, such as date of hire and so on.
Then, each of your subtables can have an EmployeeId that is both a primary key in the table and a foreign key reference to Employees.EmployeeId. This is one way to implement a one-of relationship using relational tables.
Unfortunately, unless you use triggers, this mechanism doesn't prevent one employee from being in the two other tables, but that is not part of your question.
It sounds like your design is wrong.
The entity is the employee
An attribute of an employee is their [current^] employment status.
Therefore, in its simplest form, you need a single employee table, with a column that indicates their status.
To improve this further, the employee status column should have a foreign key relationship with another table that stores the possible employee statuses.
^ current status is a 1:1 relationship. If you wanted the history of changes, this is a 1:M and needs modelling differently.
I have a major problem with my SQL design.
I need to create a database which models the following situation:
I have a list of capitals, it looks like this: Berlin, London, Prague, you get it. The other list is a list of travellers: John, Kyle, Peter etc.
The database needs to be able to answer queries such as: List of cities a given Traveller has visited, what Travellers has visited a given City and so on.
My problem is that when I create the two tables I just freeze and am unable to figure out the connection between them that would allow me to implement the intended behaviour.
I read up on it on the internet and I was told to use intersection entities, which I can, but I just don't see how that would help me. Sorry if this is a FAQ, but I just could not get my head around the proper keywords for a search.
Isn't it easier to create third table like travelers_cities with to foreign keys traveler and city, than you jan join that table with table you need and look for result there?
Solution:
Follow the following schema
First Table: Capital
Let say two columns: cid (Primary Key), Name
Second Table: Travellers
Let say two columns: tid (Primary Key), Traveller_Name
Now there is a many to many relationship that one traveller can travel/visit one or many capitals and one capital can be visited by one or many visitors.
Many to many relationship can be acheived by creating a new table which will act as reference/mapping table. let say "capital_travellers"
So, This third table will have following columns
capital_travellers:
id (Primary key): Its optional
cid (Primary key of Capital Table will work as Foreign key)
tid (Primary key of traveller Table will work as Foreign key)
Now when you want to fetch records, you will look into this table(capital_travellers).
I hope it helps.
In a many to many relationship it is necessary to implement a third junction table between the two entities. We could, say, name it travel. This table uses two foreign keys, one from each entity, to form one compound key.
Now you have 3 tables. One table called 'Traveller', one called called 'City' and a third junction table called 'Travel'. Lets assume the primary key for traveller is travellerId and for city it's cityId. The junction table takes these two keys to form a primary key for the entity 'Travel'. When u query the database for which cities a traveller with travelId '031' has travelled to it would make use of this junction table and return the cityId's to you.
If this doesn't help or if you need more clarification I recommend searching these terms:
Many-to-many
Cardinality
Given a table (Contacts) which could apply to distinct items in a database (Employers, Churches, Hospitals, Government Groups, etc.) which are stored in different tables, when leveraging this single contacts table in the end I've found there exist two choices for relating a contact back to one particular "item"
One column for each "item" type with a Foreign Key association, this results in a table looking like:
contactID empID churchID hospID govID conFN conLN ...
One column indicating the type of "item" (fkName) and one column for the value corresponding to the item of that type (fkValue). This results in a table looking like:
contactID fkName fkValue conFN conLN ...
The first means that out of the X possible foreign keys, X-1 will be NULL, but I get the advantages of hard-associated foreign keys.
The second means that I can set fkName and fkValue as NOT NULL but I don't get the advantages of DB-supported foreign keys.
Ultimately, is there a "right" answer? Are there other advantages / disadvantages that I haven't thought about (performance, security, growth/expansion)?
The second approach is an anti-pattern.
You need to set up many-to-many relationship tables between each entity (Hospitals, Churches, Employers, Government Groups, etc.) and Contacts.
If you want to make it easier to query for all of the entities a contact is related to, consider creating a view on top of the many-to-many relationship tables.
I think the second option is better as it will allow you to maintain referential integrity of your database using the in-built SQL features (foreign keys), rather than relying on your code to maintain it.
This is the solution that you should be going towards:
type
----------------
typeId name
1 hospital
2 church
contact
-----------------------------------------
contactId firstName LastName typeId (fk)
1 bob is 1
2 your uncle 2
If Bob can be a contact for more than one type, than you will need a junction table.
I have the following situation. My table is:
Table: CompanyEmployees
EmployeeID
Date of Birth
Date Joined
I also want to store the sales information for each employee. I have this:
Table: DealsCompleted
ID
EmployeeID
Deal Name
Deal Amount
My question is this- should there be a column in CompanyEmployees called "DealsCompletedID" which directly refers to the ID column in DealsCompleted, or is it acceptabe to just create a foreign key between the two Employee ID columns? Does this disadvantage the design or potentially distort the normalization?
I am unclear what the rule is as to whether I should include an extra column in CompanyEmployees or not.
EDIT Please assume there will only be one row in the deal table, per employee.
A FOREIGN KEY should point from one table to its referenced row in a parent table, the two tables should generally not reference each other (with foreign keys defined in both).
The FOREIGN KEY is most appropriately defined in the DealsCompleted table, which points back to CompanyEmployees.EmployeeID. Think about it this way - The CompanyEmployees table stores information about employees. Deals they completed do not really count as information about employees. However, the employee who completed a deal is a part of the information about a deal, so the key belongs there.
Having DealsCompleted.EmployeeID will allow for a proper one to many relationship between employees and deals. That is, one employee can have as many related rows in DealsCompleted as needed. Including a DealsCompleted column in the CompanyEmployees table on the other hand, would require you to either duplicate rows about employees, which breaks normalization, or include multiple DealCompletedID values in one column which is also incorrect.
Update after edit above Even if you plan for only a one-to-one relationship (one deal per employee), it is still more appropriate to reference the EmployeeID in DealsCompleted rather than the other way around (or both ways). ...And it allows you to expand to one-to-many, when the need arises.
Assuming that the relationship will always be one-to-one, as you state, then the answer depends on what is the primary entity within the Domain Model. If this database is at its core a database about Deals, and employee data is ancillary, then I would add an EmployeeId FK column in the Deal table. If otoh, this is a database about Employees, and Deals are ancillary, then eliminate the EmployeeId column in the Deal table, and add a DealId FK column to the Employeee table.
I read here some reasons to use constraints instead of triggers. But I have a doubt. How can be assure (using only constraints), the coherence between SUPERCLASS tables and SUBCLASSES tables?
Whit a trigger is only a matter of check when INS.. UPD...
Is there a way to define that kinda relation by using only constraints (I'm newbie at this), thanks!
You can use constraints to ensure that every ContractEmployees row has a corresponding Employees row, and likewise for SalariedExployees. I don't know of a way to use constraints to enforce the opposite: making sure that for every Employees row, there is a row either in ContractEmployees or SalariedEmployees.
Backing up a bit... there are three main ways to model OO inheritance in a relational DB. The terminology is from Martin Fowler's Patterns of Enterprise Application Architecture:
Single table inheritance: everything is just in one big table, with lots of optional columns that apply only to certain subclasses. Easy to do but not very elegant.
Concrete table inheritance: one table for each concrete type. So if all employees are either salaried or contract, you'd have two tables: SalariedEmployees and ContractEmployees. I don't like this approach either, since it makes it harder to query all employees regardless of type.
Class table inheritance: one table for the base class and one per subclass. So three tables: Employees, SalariedEmployeees, and ContractEmployees.
Here is an example of class table inheritance with constraints (code for MS SQL Server):
CREATE TABLE Employees
(
ID INT IDENTITY(1,1) NOT NULL PRIMARY KEY,
FirstName VARCHAR(100) NOT NULL DEFAULT '',
LastName VARCHAR(100) NOT NULL DEFAULT ''
);
CREATE TABLE SalariedEmployees
(
ID INT NOT NULL PRIMARY KEY REFERENCES Employees(ID),
Salary DECIMAL(12,2) NOT NULL
);
CREATE TABLE ContractEmployees
(
ID INT NOT NULL PRIMARY KEY REFERENCES Employees(ID),
HourlyRate DECIMAL(12,2) NOT NULL
);
The "REFERENCES Employees(ID)" part on the two subclass tables defines a foreign key constraint. This ensures that there must be a row in Employees for every row in SalariedEmployees or ContractEmployees.
The ID column is what links everything together. In the subclass tables, the ID is both a primary key for that table, and a foreign key pointing at the base class table.
Here's how I'd model a contract vs salary employee setup:
EMPLOYEE_TYPE_CODE table
EMPLOYEE_TYPE_CODE, pk
DESCRIPTION
Examples:
EMPLOYEE_TYPE_CODE DESCRIPTION
-----------------------------------
CONTRACT Contractor
SALARY Salaried
WAGE_SLAVE I can't be fired - slaves are sold
EMPLOYEES table
EMPLOYEE_ID, pk
EMPLOYEE_TYPE_CODE, foreign key to the EMPLOYEE_TYPE_CODE table
firstname, lastname, etc..
If you're wanting to store a hierarchical relationship, say between employee and manager (who by definition is also an employee):
EMPLOYEES table
EMPLOYEE_ID, pk
EMPLOYEE_TYPE_CODE, foreign key to the EMPLOYEE_TYPE_CODE table
MANAGER_ID
The MANAGER_ID would be filled with the employee_id of the employee who is their manager. This setup assumes that an employee could only have one manager. If you worked in a place like what you see in the movie "Office Space", you need a different setup to allow for an employee record to associate with 2+ managers:
MANAGE_EMPLOYEES_XREF table
MANAGER_EMPLOYEE_ID, pk, fk to EMPLOYEES table
EMPLOYEE_ID, pk, fk to EMPLOYEES table
Databases are relational and constraints enforce relational dependencies pretty well, been doing so for some 30 years now. What is this super and sub class you talk about?
Update
Introducing the OO inheritance relationships in databases is actually quite problematic. To take your example, contract-employee and fulltime-employee. You can model this as 1) a single table with a discriminator field, as 2) two unrelated tables, or as 3) three tables (one with the common parts, one with contract specific info, one with fulltime specific info).
However if you approach the very same problem from a traditional normal form point of view, you may end up with a structure similar to 1) or 3), but never as 2). More often than not, you'll end up with something that looks like nothing you'd recommend from your OO design board.
The problem is that when this collision of requirements happens, today almost invariably the OO design will prevail. Often times, the relational model will not even be be on the table. Why I see this as a 'problem' is that most times databases far outlive their original application. All too often I see some design that can be traced back to a OO domain driven design session from an application long forgotten, and one can see in the database schema the places where, over time, the OO design was 'hammered' into place to fit what the relational engine underneath could support, scale and deliver. The tell sign for me is tables organized on a clustered index around a identity ID when no one ever is interrogating those tables for a specific ID.