I have a column that sometimes will be null. This column is also a foreign key, so I want to know if I'll have problems with performance or with data consistency if this column will have weight
I know its a foolish question but I want to be sure.
There is no problem necessarily with this, other than it is likely indication that you might have poorly normalized design. There might be performance implications due to the way indexes are structured and the sparseness of the column with nulls, but without knowing your structure or intended querying scenarios any conclusions one might draw would be pure speculation.
A better solution might be a shared primary key where table A has a primary key, and there is zero or one records in B with the same primary key.
If table A can have one or zero B, but more than one A can refer to B, then what you have is a one to many relationship. This can be represented as Pieter laid out in his answer. This allows multiple A records to refer to the same B, and in turn each B may optionally refer to an A.
So you see there are two optional structures to address this problem, and choosing each is not guesswork. There is a distinct rational between why you would choose one or the other, but it depends on the nature of your relationships you are modelling.
Instead of this design:
create table Master (
ID int identity not null primary key,
DetailID int null references Detail(ID)
)
go
create table Detail (
ID int identity not null primary key
)
go
consider this instead
create table Master (
ID int identity not null primary key
)
go
create table Detail (
ID int identity not null primary key,
MasterID int not null references Master(ID)
)
go
Now the Foreign Key is never null, rather the existence (or not) of the Detail record indicates whether it exists.
If a Detail can exist for multiple records, create a mapping table to manage the relationship.
Related
As it is said in the title, my question is can I use int identity(1,1) for primary key in more than one table in the same ER model? I found on Internet that Primary Key need to have unique value and row, for example if I set int identity (1,1) for table:
CREATE TABLE dbo.Persons
(
Personid int IDENTITY(1,1) PRIMARY KEY,
LastName varchar(255) NOT NULL,
FirstName varchar(255),
Age int
);
GO
and the other table
CREATE TABLE dbo.Job
(
jobID int IDENTITY(1,1) NOT NULL PRIMARY KEY,
nameJob NVARCHAR(25) NOT NULL,
Personid int FOREIGN KEY REFERENCES dbo.Persons(Personid)
);
Wouldn't Personid and jobID have the same value and because of that cause an error?
Constraints in general are defined and have a scope of one table (object) in the database. The only exception is the FOREIGN KEY which usually has a REFERENCE to another table.
The PRIMARY KEY (or any UNIQUE key) sets a constraint only on the table it is defined on and is not affecting or is not affected by other constraints on other tables.
The PRIMARY KEY defines a column or a set of columns which can be used to uniquely identify one record in one table (and none of the columns can hold NULL, UNIQUE on the other hand allows NULLs and how it is treated might differ in different database engines).
So yes, you might have the same value for PersonID and JobID, but their meaning is different. (And to select the one unique record, you will need to tell SQL Server in which table and in which column of that table you are looking for it, this is the table list and the WHERE or JOIN conditions in the query).
The query SELECT * FROM dbo.Job WHERE JobID = 1; and SELECT * FROM dbo.Person WHERE PersonID = 1; have a different meaning even when the value you are searching for is the same.
You will define the IDENTITY on the table (the table can have only one IDENTITY column). You don't need to have an IDENTITY definition on a column to have the value 1 in it, the IDENTITY just gives you an easy way to generate unique values per table.
You can share sequences across tables by using a SEQUENCE, but that will not prevent you to manually insert the same values into multiple tables.
In short, the value stored in the column is just a value, the table name, the column name and the business rules and roles will give it a meaning.
To the notion "every table needs to have a PRIMARY KEY and IDENTITY, I would like to add, that in most cases there are multiple (independent) keys in the table. Usually every entity has something what you can call business key, which is in loose terms the key what the business (humans) use to identify something. This key has very similar, but usually the same characteristics as a PRIMARY KEY with IDENTITY.
This can be a product's barcode, or the employee's ID card number, or something what is generated in another system (say HR) or a code which is assigned to a customer or partner.
These business keys are useful for humans, but not always useful for computers, but they could serve as PRIMARY KEY.
In databases we (the developers, architects) like simplicity and a business key can be very complex (in computer terms), can consist of multiple columns, and can also cause performance issues (comparing a strings is not the same as comparing numbers, comparing multiple columns is less efficient than comparing one column), but the worst, it might change over time. To resolve this, we tend to create our own technical key which then can be used by computers more easily and we have more control over it, so we use things like IDENTITYs and GUIDs and whatnot.
Consider the problem above where the 'CommonChild' entity can be a child of either sub-type A or B, but not C. How would I go about designing the physical model in a relational [SQL] database?
Ideally, the solution would allow...
for an identifying relationship between CommonChild and it's related sub-type.
a 1:N relationship.
Possible Solutions
Add an additional sub-type to the super-type and move sub-type A and B under the new sub-type. The CommonChild can then have a FK constraint on the newly created sub-type. Works for the above, but not if an additional entity is added which can have a relationship with sub-type A and C, but not B.
Add a FK constraint between the CommonChild and SuperType. Use a trigger or check constraint (w/ UDF) against the super-type's discriminator before allowing a new tuple into CommonChild. Seems straight forward, but now CommonChild almost seems like new subtype itself (which it is not).
My model is fundamentally flawed. Remodel and the problem should go away.
I'm looking for other possible solutions or confirmation of one of the above solutions I've already proposed.
Thanks!
EDIT
I'm going to implement the exclusive foreign key solution provided by Branko Dimitrijevic (see accepted answer).
I am going to make a slight modifications in this case as:
the super-type, sub-type, and "CommonChild" all have the same PKs and;
the PKs are 3 column composites.
The modification is to to create an intermediate table whose sole role is to enforce the exclusive FK constraint between the sub-types and the "CommonChild" (exact model provided by Dimitrijevic minus the "CommonChild's" attributes.). The CommonChild's PK will have a normal FK constraint to the intermediate table.
This will prevent the "CommonChild" from having 2 sets of 3 column composite FKs. Plus, since the identifying relationship is maintained from super-type to "CommonChild", [read] queries can effectively ignore the intermediate table altogether.
Looks like you need a variation of exclusive foreign keys:
CREATE TABLE CommonChild (
Id AS COALESCE(SubTypeAId, SubTypeBId) PERSISTED PRIMARY KEY,
SubTypeAId int REFERENCES SubTypeA (SuperId),
SubTypeBId int REFERENCES SubTypeB (SuperId),
Attr6 varchar,
CHECK (
(SubTypeAId IS NOT NULL AND SubTypeBId IS NULL)
OR (SubTypeAId IS NULL AND SubTypeBId IS NOT NULL)
)
);
There are couple of thing to note here:
There are two NULL-able FOREIGN KEYs.
There is a CHECK that allows exactly one of these FKs be non-NULL.
There is a computed column Id which equals one of the FKs (whichever is currently non-NULL) which is also a PRIMARY KEY. This ensures that:
One parent cannot have multiple children.
A "grandchild" table can reference the CommonChild.Id directly from its FK. The SuperType.Id is effectively popagated all the way down.
We don't have to mess with NULL-able UNIQUE constraints, which are problematic in MS SQL Server (see below).
A DBMS-agnostic way of of doing something similar would be...
CREATE TABLE CommonChild (
Id int PRIMARY KEY,
SubTypeAId int UNIQUE REFERENCES SubTypeA (SuperId),
SubTypeBId int UNIQUE REFERENCES SubTypeB (SuperId),
Attr6 varchar,
CHECK (
(SubTypeAId IS NOT NULL AND SubTypeAId = Id AND SubTypeBId IS NULL)
OR (SubTypeAId IS NULL AND SubTypeBId IS NOT NULL AND SubTypeBId = Id)
)
)
Unfortunately a UNIQUE column containing more than one NULL is not allowed by MS SQL Server, which is not the case in most DBMSes. However, you can just omit the UNIQUE constraint if you don't want to reference SubTypeAId or SubTypeBId directly.
Wondering what am I missing here?
Admittedly, it is hard without having the wording of the specific problem, but things do feel a bit upside-down.
There are differences between distinct tables and type columns in terms of Performance or Optimizations for queries?
for example:
Create Table AllInOne(
Key Integer Identity Primary Key,
Desc varchar(20) Not Null,
OneType Integer Not Null
)
Where OneType only receives 1,2 or 3. (integer values)
Versus the following architecture:
Create Table One(
Key Integer Identity Primary Key,
Desc varchar(20) Not Null
)
Create Table Two(
Key Integer Identity Primary Key,
Desc varchar(20) Not Null
)
Create Table Three(
Key Integer Identity Primary Key,
Desc varchar(20) Not Null
)
Another possible architecture:
Create Table Root(
Key Integer Identity Primary Key,
Desc varchar(20) Not Null
)
Create Table One(
Key Integer Primary Key references Root
)
Create Table Two(
Key Integer Primary Key references Root
)
Create Table Three(
Key Integer Primary Key references Root
)
In the 3rd way all data will be set in the root and the relationship with the one, two and three tables.
I asked my teacher sometime ago and he couldn't answer if there is any difference.
Let's suppose i have to choose between these three approaches.
Assume that commonly used queries are filtering the type. And there are no child tables that reference these.
To make it easier to understand let's think about an payroll system.
One = Incomings
Two = Discounts
Three = Base for calculation.
Having separate tables, like in (2), will mean that someone who needs to access data for a particular OneType can ignore data for other types, thereby doing less I/O for a table scan. Also, indexes on the table in (2) would be smaller and potentially of less height, meaning less I/Os for index accesses.
Given the high selectivity of OneType, indexes would not help filtering in (1). However, table partitioning could be used to get all the benefits mentioned above.
There would also be an additional benefits. When querying (2), you need to know which OneType you need in order to know which table to query. In a partitioned version of (1), partition elimination for unneeded partitions can happen through values supplied in a where clause predicate, making the process much easier.
Other benefits include easier database management (when you add a column to a partitioned table, it gets added to all partitions), ans easier scaling (adding partitions for new OneType values is easy). Also, as mentioned, the table can be targeted by foreign keys.
I'd like to hear your suggestions on this very basic question:
Imagine these three tables:
--DROP TABLE a_to_b;
--DROP TABLE a;
--DROP TABLE b;
CREATE TABLE A
(
ID NUMBER NOT NULL ,
NAME VARCHAR2(20) NOT NULL ,
CONSTRAINT A_PK PRIMARY KEY ( ID ) ENABLE
);
CREATE TABLE B
(
ID NUMBER NOT NULL ,
NAME VARCHAR2(20) NOT NULL ,
CONSTRAINT B_PK PRIMARY KEY ( ID ) ENABLE
);
CREATE TABLE A_TO_B
(
id NUMBER NOT NULL,
a_id NUMBER NOT NULL,
b_id NUMBER NOT NULL,
somevalue1 VARCHAR2(20) NOT NULL,
somevalue2 VARCHAR2(20) NOT NULL,
somevalue3 VARCHAR2(20) NOT NULL
) ;
How would you design table a_to_b?
I'll give some discussion starters:
synthetic id-PK column or combined a_id,b_id-PK (dropping the "id" column)
When synthetic: What other indices/constraints?
When combined: Also index on b_id? Or even b_id,a_id (don't think so)?
Also combined when these entries are referenced themselves?
Also combined when these entries perhaps are referenced themselves in the future?
Heap or Index-organized table
Always or only up to x "somevalue"-columns?
I know that the decision for one of the designs is closely related to the question how the table will be used (read/write ratio, density, etc.), but perhaps we get a 20/80 solution as blueprint for future readers.
I'm looking forward to your ideas!
Blama
I have always made the PK be the combination of the two FKs, a_id and b_id in your example. Adding a synthetic id field to this table does no good, since you never end up looking for a row based on a knowledge of its id.
Using the compound PK gives you a constraint that prevents the same instance of the relationship between a and b from being inserted twice. If duplicate entries need to be permitted, there's something wrong with your data model at the conceptual level.
The index you get behind the scenes (for every DBMS I know of) will be useful to speed up common joins. An extra index on b_id is sometimes useful, depending on the kinds of joins you do frequently.
Just as a side note, I don't use the name "id" for all my synthetic pk columns. I prefer a_id, b_id. It makes it easier to manage the metadata, even though it's a little extra typing.
CREATE TABLE A_TO_B
(
a_id NUMBER NOT NULL REFERENCES A (a_id),
b_id NUMBER NOT NULL REFERENCES B (b_id),
PRIMARY KEY (a_id, b_id),
...
) ;
It's not unusual for ORMs to require (or, in more clueful ORMs, hope for) an integer column named "id" in addition to whatever other keys you have. Apart from that, there's no need for it. An id number like that makes the table wider (which usually degrades I/O performance just slightly), and adds an index that is, strictly speaking, unnecessary. It isn't necessary to identify the entity--the existing key does that--and it leads new developers into bad habits. (Specifically, giving every table an integer column named "id", and believing that that column alone is the only key you need.)
You're likely to need one or more of these indexed.
a_id
b_id
{a_id, b_id}
{b_id, a_id}
I believe Oracle should automatically index {a_id, b_id}, because that's the primary key. Oracle doesn't automatically index foreign keys. Oracle's indexing guidelines are online.
In general, you need to think carefully about whether you need ON UPDATE CASCADE or ON DELETE CASCADE. In Oracle, you only need to think carefully about whether you need ON DELETE CASCADE. (Oracle doesn't support ON UPDATE CASCADE.)
the other comments so far are good.
also consider adding begin_dt and end_dt to the relationship. in this way, you can manage a good number of questions about each relationship through time. (consider baseline issues)
Just trying to figure out the best way to design my table for the following scenario:
I have several areas in my system (documents, projects, groups and clients) and each of these can have comments logged against them.
My question is should I have one table like this:
CommentID
DocumentID
ProjectID
GroupID
ClientID
etc
Where only one of the ids will have data and the rest will be NULL or should I have a separate CommentType table and have my comments table like this:
CommentID
CommentTypeID
ResourceID (this being the id of the project/doc/client)
etc
My thoughts are that option 2 would be more efficient from an indexing point of view. Is this correct?
Option 2 is not a good solution for a relational database. It's called polymorphic associations (as mentioned by #Daniel Vassallo) and it breaks the fundamental definition of a relation.
For example, suppose you have a ResourceId of 1234 on two different rows. Do these represent the same resource? It depends on whether the CommentTypeId is the same on these two rows. This violates the concept of a type in a relation. See SQL and Relational Theory by C. J. Date for more details.
Another clue that it's a broken design is that you can't declare a foreign key constraint for ResourceId, because it could point to any of several tables. If you try to enforce referential integrity using triggers or something, you find yourself rewriting the trigger every time you add a new type of commentable resource.
I would solve this with the solution that #mdma briefly mentions (but then ignores):
CREATE TABLE Commentable (
ResourceId INT NOT NULL IDENTITY,
ResourceType INT NOT NULL,
PRIMARY KEY (ResourceId, ResourceType)
);
CREATE TABLE Documents (
ResourceId INT NOT NULL,
ResourceType INT NOT NULL CHECK (ResourceType = 1),
FOREIGN KEY (ResourceId, ResourceType) REFERENCES Commentable
);
CREATE TABLE Projects (
ResourceId INT NOT NULL,
ResourceType INT NOT NULL CHECK (ResourceType = 2),
FOREIGN KEY (ResourceId, ResourceType) REFERENCES Commentable
);
Now each resource type has its own table, but the serial primary key is allocated uniquely by Commentable. A given primary key value can be used only by one resource type.
CREATE TABLE Comments (
CommentId INT IDENTITY PRIMARY KEY,
ResourceId INT NOT NULL,
ResourceType INT NOT NULL,
FOREIGN KEY (ResourceId, ResourceType) REFERENCES Commentable
);
Now Comments reference Commentable resources, with referential integrity enforced. A given comment can reference only one resource type. There's no possibility of anomalies or conflicting resource ids.
I cover more about polymorphic associations in my presentation Practical Object-Oriented Models in SQL and my book SQL Antipatterns.
Read up on database normalization.
Nulls in the way you describe would be a big indication that the database isn't designed properly.
You need to split up all your tables so that the data held in them is fully normalized, this will save you a lot of time further down the line guaranteed, and it's a lot better practice to get into the habit of.
From a foreign key perspective, the first example is better because you can have multiple foreign key constraints on a column but the data has to exist in all those references. It's also more flexible if the business rules change.
To continue from #OMG Ponies' answer, what you describe in the second example is called a Polymorphic Association, where the foreign key ResourceID may reference rows in more than one table. However in SQL databases, a foreign key constraint can only reference exactly one table. The database cannot enforce the foreign key according to the value in CommentTypeID.
You may be interested in checking out the following Stack Overflow post for one solution to tackle this problem:
MySQL - Conditional Foreign Key Constraints
The first approach is not great, since it is quite denormalized. Each time you add a new entity type, you need to update the table. You may be better off making this an attribute of document - I.e. store the comment inline in the document table.
For the ResourceID approach to work with referential integrity, you will need to have a Resource table, and a ResourceID foreign key in all of your Document, Project etc.. entities (or use a mapping table.) Making "ResourceID" a jack-of-all-trades, that can be a documentID, projectID etc.. is not a good solution since it cannot be used for sensible indexing or foreign key constraint.
To normalize, you need to the comment table into one table per resource type.
Comment
-------
CommentID
CommentText
...etc
DocumentComment
---------------
DocumentID
CommentID
ProjectComment
--------------
ProjectID
CommentID
If only one comment is allowed, then you add a unique constraint on the foreign key for the entity (DocumentID, ProjectID etc.) This ensures that there can only be one row for the given item and so only one comment. You can also ensure that comments are not shared by using a unique constraint on CommentID.
EDIT: Interestingly, this is almost parallel to the normalized implementation of ResourceID - replace "Comment" in the table name, with "Resource" and change "CommentID" to "ResourceID" and you have the structure needed to associate a ResourceID with each resource. You can then use a single table "ResourceComment".
If there are going to be other entities that are associated with any type of resource (e.g. audit details, access rights, etc..), then using the resource mapping tables is the way to go, since it will allow you to add normalized comments and any other resource related entities.
I wouldn't go with either of those solutions. Depending on some of the specifics of your requirements you could go with a super-type table:
CREATE TABLE Commentable_Items (
commentable_item_id INT NOT NULL,
CONSTRAINT PK_Commentable_Items PRIMARY KEY CLUSTERED (commentable_item_id))
GO
CREATE TABLE Projects (
commentable_item_id INT NOT NULL,
... (other project columns)
CONSTRAINT PK_Projects PRIMARY KEY CLUSTERED (commentable_item_id))
GO
CREATE TABLE Documents (
commentable_item_id INT NOT NULL,
... (other document columns)
CONSTRAINT PK_Documents PRIMARY KEY CLUSTERED (commentable_item_id))
GO
If the each item can only have one comment and comments are not shared (i.e. a comment can only belong to one entity) then you could just put the comments in the Commentable_Items table. Otherwise you could link the comments off of that table with a foreign key.
I don't like this approach very much in your specific case though, because "having comments" isn't enough to put items together like that in my mind.
I would probably go with separate Comments tables (assuming that you can have multiple comments per item - otherwise just put them in your base tables). If a comment can be shared between multiple entity types (i.e., a document and a project can share the same comment) then have a central Comments table and multiple entity-comment relationship tables:
CREATE TABLE Comments (
comment_id INT NOT NULL,
comment_text NVARCHAR(MAX) NOT NULL,
CONSTRAINT PK_Comments PRIMARY KEY CLUSTERED (comment_id))
GO
CREATE TABLE Document_Comments (
document_id INT NOT NULL,
comment_id INT NOT NULL,
CONSTRAINT PK_Document_Comments PRIMARY KEY CLUSTERED (document_id, comment_id))
GO
CREATE TABLE Project_Comments (
project_id INT NOT NULL,
comment_id INT NOT NULL,
CONSTRAINT PK_Project_Comments PRIMARY KEY CLUSTERED (project_id, comment_id))
GO
If you want to constrain comments to a single document (for example) then you could add a unique index (or change the primary key) on the comment_id within that linking table.
It's all of these "little" decisions that will affect the specific PKs and FKs. I like this approach because each table is clear on what it is. In databases that's usually better then having "generic" tables/solutions.
Of the options you give, I would go for number 2.
Option 2 is a good way to go. The issue that I see with that is you are putting the resouce key on that table. Each of the IDs from the different resources could be duplicated. When you join resources to the comments you will more than likely come up with comments that do not belong to that particular resouce. This would be considered a many to many join. I would think a better option would be to have your resource tables, the comments table, and then tables that cross reference the resource type and the comments table.
If you carry the same sort of data about all comments regardless of what they are comments about, I'd vote against creating multiple comment tables. Maybe a comment is just "thing it's about" and text, but if you don't have other data now, it's likely you will: date the comment was entered, user id of person who made it, etc. With multiple tables, you have to repeat all these column definitions for each table.
As noted, using a single reference field means that you could not put a foreign key constraint on it. This is too bad, but it doesn't break anything, it just means you have to do the validation with a trigger or in code. More seriously, joins get difficult. You can just say "from comment join document using (documentid)". You need a complex join based on the value of the type field.
So while the multiple pointer fields is ugly, I tend to think that's the right way to go. I know some db people say there should never be a null field in a table, that you should always break it off into another table to prevent that from happening, but I fail to see any real advantage to following this rule.
Personally I'd be open to hearing further discussion on pros and cons.
Pawnshop Application:
I have separate tables for Loan, Purchase, Inventory & Sales transactions.
Each tables rows are joined to their respective customer rows by:
customer.pk [serial] = loan.fk [integer];
= purchase.fk [integer];
= inventory.fk [integer];
= sale.fk [integer];
I have consolidated the four tables into one table called "transaction", where a column:
transaction.trx_type char(1) {L=Loan, P=Purchase, I=Inventory, S=Sale}
Scenario:
A customer initially pawns merchandise, makes a couple of interest payments, then decides he wants to sell the merchandise to the pawnshop, who then places merchandise in Inventory and eventually sells it to another customer.
I designed a generic transaction table where for example:
transaction.main_amount DECIMAL(7,2)
in a loan transaction holds the pawn amount,
in a purchase holds the purchase price,
in inventory and sale holds sale price.
This is clearly a denormalized design, but has made programming alot easier and improved performance. Any type of transaction can now be performed from within one screen, without the need to change to different tables.