I am a newcomer to SQL Server undertaking a migration of a web site that I have written from Access 2003 to SQL Server, using ASP. I have spent a lot of time analysing the Access databases. I have also done a lot of reading, but can't find an answer to these questions:

If a table has a compound Primary Key (up to 4 rows) can you
a) make the whole primary key a clustered index?
b) make part of the key a clustered index (1 row only)
c) is there a single index for the primary key, and if so can I separately index parts of the primary key?

I appreciate how important it is to get the indexing right and I want to be absolutely clear on what I can and can't do.

Thank you
Pat Waddington

-------------------
System 3 2000 Limited
www.system3-2000.co.uk

__________________
-------------------
System 3 2000 Limited
www.system3-2000.co.uk (but very out-of-date!)

a) Yes, for example the OrderDetails table from Northwind db

ALTER TABLE [dbo].[Order Details] WITH NOCHECK ADD
    CONSTRAINT [PK_Order_Details] PRIMARY KEY CLUSTERED
    (
        [OrderID],
        [ProductID]
    ) ON [PRIMARY]
GO
b) NO because clustered index must be unique
c) Yes there is a single index, and Yes you can create additional indexes which only cover part of the primary key. Not sure why you'd want to though...have you noticed that SQL Server primary keys are much more flexible than Access? You don't have to have the 1st fields in the table in that order in the PK. You can have any fields in any order. You should put the most selective fields first when you create the PK (e.g. a unique value is highly selective, but a value found in half the rows is not).


This is a good article, and a good site
http://www.sql-server-performance.co..._not_equal.asp

hth
Phil