Uniqueidentifier vs. int identity(1,1): Discuss.

So I'm being thrown at a Microsoft Dynamics CRM application, and I see that all the PK / FK's are uniqueidentifier, and not the int identity(1,1) I've seen on almost every other application I've worked with.

Question:  What is the advantage of uniqueidentifier, as opposed to int identity(1,1) ?

Thanks in advance.
Jim
LVL 66
Jim HornMicrosoft SQL Server Developer, Architect, and AuthorAsked:
Who is Participating?
 
Scott PletcherSenior DBACommented:
The use of all GUIDs and/or IDENTITYs is a lazy dodge to proper table designs.

A GUID is worse, of course, because of its randomness and extreme length.

But, contrary to general assumptions, often a natural key of the proper type will perform vastly better on joins than a meaningless identity.  Parent/child joins can be especially faster when clustering the child rows first on the parent's key.
0
 
AshokCommented:
One advantage a GUID has is that it tends to be unique even across databases, but it's a performance and space hit in most applications.

int identity(1,1) uses less space and sequential so it is faster.
0
 
Aneesh RetnakaranDatabase AdministratorCommented:
Only advantage occurs when you insert data from different servers / databases; as the uniqueidentifier is unique, there wont be any issues while inserting the data; other than this, it makes the database larger.  (16bytes  vs 4 bytes , that means per row 12 bytes extra for unique identifier  )
0
Introducing Cloud Class® training courses

Tech changes fast. You can learn faster. That’s why we’re bringing professional training courses to Experts Exchange. With a subscription, you can access all the Cloud Class® courses to expand your education, prep for certifications, and get top-notch instructions.

 
AshokCommented:
0
 
AshokCommented:
Identity does an auto update whereas uniqueidentifier is data specific.
0
 
Kent OlsenData Warehouse Architect / DBACommented:
Another drawback to unique vs. identity is performance.  

The most efficient join key is an integer.  The least efficient is a string.  On large tables, the joins will be noticeably slower on a unique identifier than on an integer join.
0
 
Kent OlsenData Warehouse Architect / DBACommented:
>> The use of all GUIDs and/or IDENTITYs is a lazy dodge to proper table designs.

That's not really true.  The data can be organized and stored to be most advantageous.  An identity just enumerates the data and becomes the fastest join key.  Dimension tables in a star schema are a good example.


Kent
0
 
Scott PletcherSenior DBACommented:
We're talking OLTP not OLAP.

I think it's lazy to assume that the proper clustering key for ALL data at ALL times is a meaningless, auto-generated number.  I'm glad the Contacts on my phone are ordered by Name and not by the order I added them.
0
 
Kent OlsenData Warehouse Architect / DBACommented:
If your phone book is joined to other demographics, the join will be faster if there's an identity column in the phone book and the join is performed on the scalar integer identity key than on the composite string key that is last, first.
0
 
Jim HornMicrosoft SQL Server Developer, Architect, and AuthorAuthor Commented:
Thanks all.  Points awarded based on perspective that I counldn't find with a standard Google search.  -Jim
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.