asked on

Dynamic SQL in Function UDF

Hi,

following problem, SQL Server 2005:

I have a user definded function (needs to be a function because i use it within a query) that returns a value such like this:

CREATE FUNCTION [dbo].[fct_prg_GetCostsPerProject](@Project_ID int)

	RETURNS money
AS
BEGIN
	DECLARE @Result money

	SELECT     @Result=ISNULL(SUM(Costs),0)
	FROM         dbo.myTable
	WHERE     Mx_Project_fID = @Project_ID

		
	return @Result
END

Open in new window

My problem now is that the name of the table (here: myTable) depends on the variable @Project_ID that is passed to the function (for instance if @Project_ID=5 the table name would be somothing like myTable_5)

I know that dynamic SQL is not working withing UDF's but has anyone an idea for a workaround?

Temporary tables are no solution because the table handles huge amount of records

Thanks in advance
Andy

Guy Hengel [angelIII / a3]

the only "workaround" is to put the IF @projectid = '5' then use table mytable_5 ...
but that is very far from "dynamic" and "manageable"

sorry, but there is no solution for this.

cas1

ASKER

Hi, angellll,

unfortunately this is impossible to manage due to the table names are not limited to a certain numbers

Andy

Guy Hengel [angelIII / a3]

the only thing I see is that you build up a view on all the relevant tables, along with the field that you would map the projectid value to.
is the structure of those table(s) the same? in which case, why not have 1 single table (eventually partitioned) ...

anyhow, the view method would be like this:

CREATE VIEW dbo.MyView
AS
 SELECT 5 ProjectID, Costs
     FROM         dbo.myTable5
   UNION ALL
 SELECT 6 ProjectID, Costs
     FROM         dbo.myTable6

Open in new window

and your function:

CREATE FUNCTION [dbo].[fct_prg_GetCostsPerProject](@Project_ID int)

        RETURNS money
AS
BEGIN
        DECLARE @Result money

        SELECT     @Result=ISNULL(SUM(Costs),0)
        FROM         dbo.myView
        WHERE     ProjectID = @Project_ID
                
        return @Result
END

Open in new window

cas1

ASKER

My problem is: at this time that I have one table with all values in, using the @Project_ID to separate beteween the different projects. Since there are > 17.000.000 recordsets in the table (lots of projects, each project up to 800.000 recordsets) i have heavy performance issues. The table holds plenty of columns that all needs to be evaluated in different evaluations. I tried around with setting the right indexes using the execution plan of the queries and the index recommondations, but I would need to place over 15 indexes to solve the performance issuse when evaluating the table. But the disadvantage is that all operations with the table itself become very very slow (due to the number of indexes). So deleting and recreating the 800.000 records of a single project take much to long.

So my approach was to have a seperate table for each project. That would solve my performance issues.

Maybe you have a solution for this ???

Andy

ASKER CERTIFIED SOLUTION

Guy Hengel [angelIII / a3]

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

vandalesm

If you are using SQL Server 2005 or greater. I think partitioning your table is the best option.
It is very simple to setup and there are lot of information in the web.

Aaron Shilo

hi

another solution that is very easy to implement is "Filtered Indexes"
if you have 2008 version.

a index that will hold only data requierd for you in queries by filter.

cas1

ASKER

Setting the clustered index works great and prevent me from a database redesign

Thanks a lot