Hi,
What is the difference between a computed column and a UDF?
Is a computed column the same as the "Formula" field under Design Table in Enterprise Manager?
Also, what is the proper syntax for the Formula field? Can I use regular SQL on it or is there more to it?
thanks,
Frank
Is it possible to get a Computed column in one table to carry out a SELECT & SUM from a column in another table? I have tried a SELECT / FROM / WHERE construct but SQL complains about that.
Regards
Clive
I am working with data that has a lot of records that get updated and inserted frequently, and to avoid having to create formulas through code all over the place, I am experimenting with computed column formulas. I have a question though.. It works well for any addition or subtraction (columnA+columnB), however, when I try to use division (columnA/columnB), it only returns integers, no decimals. I would like to have decimals, particularly with a specific scale and precision and I would really like to attempt this without any coding. Any suggestions?
View 3 Replies View RelatedI have a table with fields called fname (First Name) and lname (Last Name). I need the userīs email thai is compose from lname and fname:
LOWER(LEFT (fname,1) + lname)
Is there any difference between creatig this computed column ia a table or in a view in SQL Server 2000?
I can do:
1. CREATE TABLE Users(
fname varchar(20),
lname varchar(20),
email as LOWER(LEFT (fname,1) + lname) )
Or
2. CREATE TABLE Users (
fname varchar(20),
lname varchar(20))
CREATE VIEW Vw_users (fname, Lname ,
email)
AS
SELECT fname, Lname ,
LOWER(LEFT (fname,1) + lname) )
Is one of them is better?
Paulo
Dear all,
Pls help me with this
I have 2 tables
Trip (TripID, Duration)
Reserve(ReserveID, TripID, StartDay, EndDate)
in which EndDate = StartDay + Trip.Duration
How can I do this?
I have a table with fields called fname (First Name) and lname (LastName). I need the userīs email thai is compose from lname and fname:LOWER(LEFT (fname,1) + lname)Is there any difference between creatig this computed column ia a tableor in a view in SQL Server 2000?I can do:1. CREATE TABLE Users(fname varchar(20),lname varchar(20),email as LOWER(LEFT (fname,1) + lname) )Or2. CREATE TABLE Users (fname varchar(20),lname varchar(20))CREATE VIEW Vw_users (fname, Lname ,email)ASSELECT fname, Lname ,LOWER(LEFT (fname,1) + lname) )Is one of them is better?Paulo*** Sent via Developersdex http://www.developersdex.com ***Don't just participate in USENET...get rewarded for it!
View 1 Replies View RelatedHi,
Consider the following example
create table sample
(col1 int,
col2 int ,
col3 AS col1 + col2) PERSISTED NOT NULL)
basically col3 is a computed column. Now when ever a row in col1 or col2 is updated the computed column will reflect the new value. how does this happen in the background. does this use row level triggers or what other mechanism is used to maintain col3 - computed column
Hi, I have a small problem with my database. I've got following situation: I have a computed column, which value is base on currency rate: rent * rate. Users have to have possibility to change currency rate easily (maybe another table or constant). Is there any way to create formula, which would compute value properly, via constant or something like this? Or the easiest workaround would be load data into dataset (I'm building asp.net application - database will be very small - couple of hundreds of records) and make calculations programmatically?
Przemek
Hi,
I have a table with 4 columns let us say A,B,C,D.
column D is computed column with formula A + '-' + B
Now, i want to add one more condition to the formula which looks like "A + '-' + B + '-' + C".
Please let me know how to do this using T-SQL as i cannot open the table in design mode in production server.
Thanks in Advance!!
i have a table containing the column "current month" and "current day" as smallint which contains the number of months since 1900-01-01 and the day of this month. now i want to trnslate this column in a smalldatetime ( not datetime !) value using a computed column and then create an index on that column.
the formula should be:
dateadd(d,[current day]-1,
dateadd(m,[current month],convert(smalldatetime,'1900-01-01'))
)
trying to create an index on this column results in an error message saying
that the formula is nondeterministic or imprecise
removing the convert statement leaving only the date results in a column of type datetime and creating the index works fine
replacing convert(smalldatetime,'1900-01-01') with a column name which has the type smalldatetime also allows to create an index but thats not what i want to do.
it seems that sql2000 thinks a convert from a string to a date is nondeterministic. Is there any possibility to create a const of type smalldatetime without using convert?
Any idea?
(besides this, datediff(d,'yyyy-mm-dd',anydate) is nondeterministic but datediff(d,dateadd(d,0,'yyyy-mm-dd'),anydate) is deterministic. strange...)
and
I am having a problem with using UDF as part of a temp table computed column. Here's the sample code:
IF EXISTS( SELECT 1 FROM information_schema.routines WHERE routine_name = 'fn_test')
DROP FUNCTION dbo.fn_testGO
CREATE FUNCTION dbo.fn_test( @x int, @y int)
RETURNS INT AS
BEGIN
DECLARE @z INT
SET @z = @x + @y
RETURN @z
END
GO
CREATE TABLE #X
(
x INT,
y INT,
z AS (dbo.fn_test(x,y))
)
I receive the following error:
Server: Msg 208, Level 16, State 1, Line 2
Invalid object name 'dbo.fn_test'.
I do not get this error if I use a regular table.
HELP!
Suppose a table being
Create table myTable (ID int, col1 int, col2 int)
I know how to make a computed column being the sum of other column for the same ID e.g. "computed_column = col1 + col2".
Getting the average would be "computed_column = (col1 + col2)/2" But how to get the Max, Min?
Even "Sum(col1,col2)" or AVG(col1, col2) does not work as the formula for a computed column...
I have a table that I cannot allow a computed field to exist on (due to a 3rd party software), so I am thinking I could create a view with a computed field that is persistent, is that possible?
the syntax below will not work, I am not even sure if this is possible, but if it can work, that would be great.
I am wanting to get the sum of jetfoot1, 2 & 3 and have the total added up as "total"
create view ViewSumReport as
select JETFOOT1,JETFOOT2,JETFOOT3,(JETFOOT1+JETFOOT2+JETFOOT3)as [total] persisted
from dbo.fielddata
GO
I have a checksum calculation as a persisted, indexed computed column on a temporary table that I used to compare against original records to detect changes.
It seems that the update/ insert statements in my procs get out of sync on larger tables (500,000 rows +) with the checksum calculations. The only thing I can think of is that the column calculations are performed asynchronously in relation to the updates/ inserts. This is a problem for me.
Is my assumption correct? If it is, how can I adjust for this, i.e., force the computations to be performed synchronously or wait for the computations to complete before running comparisons?
-Jeremy
___________________________
Geek At Large
In SQL Sever, do the size of computed columns gets added to the totalsize of the tables? Does SQL server stores the actual values incomputed columns?Thanks_GJK
View 1 Replies View RelatedI need to create an function similar to the "MATCH" function in Excelthat evaluates a number within a set of numbers and returns whetherthere is a match. I have put the example of what I see in excel in thecheck column. The "0" answer in the result column is in the fourthaccount in the list. Somehow I need to loop through the accountscomparing the result to the total and indicate a match in the checkcolumn. It wouldn't even need to tell me the row number; it could be a0 or 1.account total result check123770266.84124.2112377026131.050 412377026164.38-33.33123770260131.051237702678.7152.3412377167-31.34221.891237716731.34159.211237716738.55152 51237716731.34159.211237716715238.5512377167490.91-300.36123771670190.55123771670190.5512377167-31.3443.341237716731.34-19.341237716738.55-26.551237716731.34-19.3412377167152-14012377167490.91-478.9112377167012123771670121237736347.058412377363131.05012377363-45.38176.4312377363-47.05178.11237736347.0484.0112377363-47.04178.091237736347.058412377363541.11-410.06123773630131.0512377363672.15-541.11237750737.64152.91
View 3 Replies View RelatedI have a checksum calculation as a persisted, indexed computed column on a temporary table that I used to compare against original records to detect changes.
It seems that the update/ insert statements in my procs get out of sync on larger tables (500,000 rows +) with the checksum calculations. The only thing I can think of is that the column calculations are performed asynchronously in relation to the updates/ inserts. This is a problem for me.
Is my assumption correct? If it is, how can I adjust for this, i.e., force the computations to be performed synchronously or wait for the computations to complete before running comparisons?
Does tsql allow sth like
Select col1*col2 as ComputedColumn, ComputedColumn + 2 as NewColumn
From T_Table
THis is possible in Access.
Hello,
does anyone know a website, where I can read something about the syntax of Computed Columns?
I don't know how to enter the following expression in the computed columns field of MS SQL Server:
When x-y < 0 Then 0 else x-y
Thank you
M-l-G
Instead of using Full-Text indices, which I don't like to manage, we've tried to use seperate tables that contain recordID, the word, a count of the word in the parent field and computed column which is the CHECKSUM() of the word column. I indexed the checksum column with a clustered index.
Works great in Query Analyser. But when the ASP page calls it, I get this message:
Microsoft OLE DB Provider for SQL Server (0x80040E14)
INSERT failed because the following SET options have incorrect settings: 'ARITHABORT'.
Same for updates and deletes. The question is how should these SET settings be done? Any ideas would be greatly welcomed.
Thanks
Jason
Using SS2K, I'm getting the following error while bulk inserting:Column 'warranty_expiration_date' cannot be modified because it is acomputed column.Here is my bulk insert statement:BULK INSERT dbo.TestDataFROM 'TestData.dat'WITH (CHECK_CONSTRAINTS,FIELDTERMINATOR='|',MAXERRORS = 1,FORMATFILE='TestData.fmt')The computed column is not referenced in the format file and the data filedoes not contain the computed data.Thanks
View 2 Replies View RelatedHow would I go about tracing UDF performance in profiler? I'd like to
specifically know the impact of the UDF without having to dig into the
execution plan of the statement containing it. Is this possible?
Can changes made to replicated stored procedures (and User-defined Functions) on the Publisher update the Subscriber database?
If I ALTER a SP (or UDF) on Database1 with peer-to-peer Transaction Replication to Database2, will the alter replicate to the Subscriber(s)?
Thomas
ThomBeaux
I know there is a lot of information already out there on this topic,but given the following scenario...--------------------------------------------------------------------------------------------------------------------------------------Create a view like so ( pardon the pseudo-code )...CREATE View vwContactAddressesSelect * FROM Contact INNER JOIN Address ON Contact.ContactID =Address.ContactIDAnd then do a sargable select from the view using a stored procedureCREATE STORED PROCEDURE spSelect_ContactAddresses@ContactID intASSelect * FROM vwContactAddresses WHERE ContactID = @ContactID--------------------------------------------------------------------------------------------------------------------------------------In my understanding, "vwContactAddresses" would be substituted with theactual SQL join statement when the view is accessed.So for the stored procedure in question an execution plan for"Select * FROM Contact INNER JOIN Address ON Contact.ContactID =Address.ContactID WHERE ContactID = @ContactID" would be cached.Correct?With regards to execution plan caching, is this not the same ascreating an inline UDF that takes parameters or just creating a storedprocedure that would do the join w/out the view reference?
View 3 Replies View Related-- Author: Igor Nikiforov, Montreal, EMail: udfs@sympatico.ca
-- GETWORDCOUNT() User-Defined Function Counts the words in a string.
-- GETWORDCOUNT(@cString[, @cDelimiters])
-- Parameters
-- @cString nvarchar(4000) - Specifies the string whose words will be counted.
-- @cDelimiters nvarchar(256) - Optional. Specifies one or more optional characters used to separate words in @cString.
-- The default delimiters are space, tab, carriage return, and line feed. Note that GETWORDCOUNT( ) uses each of the characters in @cDelimiters as individual delimiters, not the entire string as a single delimiter.
-- Return Value smallint
-- Remarks GETWORDCOUNT() by default assumes that words are delimited by spaces or tabs. If you specify another character as delimiter, this function ignores spaces and tabs and uses only the specified character.
-- If you use 'AAA aaa, BBB bbb, CCC ccc.' as the target string for dbo.GETWORDCOUNT(), you can get all the following results.
-- declare @cString nvarchar(4000)
-- set @cString = 'AAA aaa, BBB bbb, CCC ccc.'
-- select dbo.GETWORDCOUNT(@cString, default) -- 6 - character groups, delimited by ' '
-- select dbo.GETWORDCOUNT(@cString, ',') -- 3 - character groups, delimited by ','
-- select dbo.GETWORDCOUNT(@cString, '.') -- 1 - character group, delimited by '.'
-- See Also GETWORDNUM() User-Defined Function
-- UDF the name and functionality of which correspond to the same built-in function of Visual FoxPro
CREATE function GETWORDCOUNT (@cSrting nvarchar(4000), @cDelimiters nvarchar(256) )
returns smallint
as
begin
-- if no break string is specified, the function uses spaces, tabs and line feed to delimit words.
set @cDelimiters = isnull(@cDelimiters, space(1)+char(9)+char(10))
declare @p smallint, @end_of_string smallint, @wordcount smallint
select @p = 1, @wordcount = 0
select @end_of_string = 1 + datalength(@cSrting)/(case SQL_VARIANT_PROPERTY(@cSrting,'BaseType') when 'nvarchar' then 2 else 1 end) -- for unicode
while dbo.CHARINDEX_BIN(substring(@cSrting, @p, 1), @cDelimiters, 1) > 0 and @end_of_string > @p -- skip opening break characters, if any
set @p = @p + 1
if @p < @end_of_string
begin
set @wordcount = 1 -- count the one we are in now count transitions from 'not in word' to 'in word'
-- if the current character is a break char, but the next one is not, we have entered a new word
while @p < @end_of_string
begin
if @p +1 < @end_of_string and dbo.CHARINDEX_BIN(substring(@cSrting, @p, 1), @cDelimiters, 1) > 0 and dbo.CHARINDEX_BIN(substring(@cSrting, @p+1, 1), @cDelimiters, 1) = 0
select @wordcount = @wordcount + 1, @p = @p + 1 -- Skip over the first character in the word. We know it cannot be a break character.
set @p = @p + 1
end
end
return @wordcount
end
GO
-- Author: Igor Nikiforov, Montreal, EMail: udfs@sympatico.ca
-- GETWORDNUM() User-Defined Function
-- Returns a specified word from a string.
-- GETWORDNUM(@cString, @nIndex[, @cDelimiters])
-- Parameters @cString nvarchar(4000) - Specifies the string to be evaluated
-- @nIndex smallint - Specifies the index position of the word to be returned. For example, if @nIndex is 3, GETWORDNUM( ) returns the third word (if @cString contains three or more words).
-- @cDelimiters nvarchar(256) - Optional. Specifies one or more optional characters used to separate words in @cString.
-- The default delimiters are space, tab, carriage return, and line feed. Note that GetWordNum( ) uses each of the characters in @cDelimiters as individual delimiters, not the entire string as a single delimiter.
-- Return Value nvarchar(4000)
-- Remarks Returns the word at the position specified by @nIndex in the target string, @cString. If @cString contains fewer than @nIndex words, GETWORDNUM( ) returns an empty string.
-- See Also
-- GETWORDCOUNT() User-Defined Function
-- UDF the name and functionality of which correspond to the same built-in function of Visual FoxPro
CREATE function GETWORDNUM (@cSrting nvarchar(4000), @nIndex smallint, @cDelimiters nvarchar(256) )
returns nvarchar(4000)
as
begin
-- if no break string is specified, the function uses spaces, tabs and line feed to delimit words.
set @cDelimiters = isnull(@cDelimiters, space(1)+char(9)+char(10))
declare @i smallint, @j smallint, @p smallint, @q smallint, @qmin smallint, @end_of_string smallint, @LenDelimiters smallint, @outstr nvarchar(4000)
select @i = 1, @p = 1, @q = 0, @outstr = ''
select @end_of_string = 1 + datalength(@cSrting)/(case SQL_VARIANT_PROPERTY(@cSrting,'BaseType') when 'nvarchar' then 2 else 1 end) -- for unicode
select @LenDelimiters = datalength(@cDelimiters)/(case SQL_VARIANT_PROPERTY(@cDelimiters,'BaseType') when 'nvarchar' then 2 else 1 end) -- for unicode
while @i <= @nIndex
begin
while dbo.CHARINDEX_BIN(substring(@cSrting, @p, 1), @cDelimiters, 1) > 0 and @end_of_string > @p -- skip opening break characters, if any
set @p = @p + 1
if @p >= @end_of_string
break
select @j = 1, @qmin = @end_of_string -- find next break character it marks the end of this word
while @j <= @LenDelimiters
begin
set @q = dbo.CHARINDEX_BIN(substring(@cDelimiters, @j, 1), @cSrting, @p)
set @j = @j + 1
if @q > 0 and @qmin > @q
set @qmin = @q
end
if @i = @nIndex -- this is the actual word we are looking for
begin
set @outstr = substring(@cSrting, @p, @qmin-@p)
break
end
set @p = @qmin + 1
if (@p >= @end_of_string)
break
set @i = @i + 1
end
return @outstr
end
GO
-- Is similar to the built-in function Transact-SQL charindex, but regardless of collation settings,
-- executes case-sensitive search
-- Author: Igor Nikiforov, Montreal, EMail: udfs@sympatico.ca
CREATE function CHARINDEX_BIN(@expression1 nvarchar(4000), @expression2 nvarchar(4000), @start_location smallint = 1)
returns nvarchar(4000)
as
begin
return charindex( cast(@expression1 as nvarchar(4000)) COLLATE Latin1_General_BIN, cast(@expression2 as nvarchar(4000)) COLLATE Latin1_General_BIN, @start_location )
end
GO
Hi!I need to refresh an entire database.I can recompile SPs with sp_recompile (or DBCC FLUSHPROCINDB), andrefresh views with sp_refreshView, but I cannot find any way torefresh my user-defined functions (some of them are like views, withparameters).Any help appreciated :) !Ben
View 5 Replies View RelatedHi,
I have written an udf that checks for validations of an email address and returns 0 / 1 as per the validations. I came across an artical on MSDN that describes the steps to make my own system udf. http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnsqlpro01/html/sql01l1.asp
with help of that artical, I made my udf as a system udf. and now I can use it as follows.
select * , fn_isvalidemail(email) from #tp
now, I wanted to use this as a declarative check constraints in my tables. so I tried
create table testtable
( email_addr varchar (255) null CONSTRAINT CK_email_chk CHECK ( fn_isvalidemail(email) = 1) )
although this is syntax is logically correct, it does not get compiled
it gives me an error saying,
Server: Msg 195, Level 15, State 10, Line 2
'fn_isvalidemail' is not a recognized function name.
the strange part is that i made the function as another normal udf, It works in declarative constraints.
create table testtable
( email varchar (255) null CONSTRAINT CK_1p12 CHECK ( dbo.isvalidemail(email) = 1) )
Any thoughts on this ?
- Amit
I have always learn that UDFs are not Pre-Compiled while Stored procedures are. But I always had this questions why are UDFs are not Pre-Compiled.
I searched online for the answer but dint get an concrete justification on the same.
Hello,
How can I call UDFs on linked servers in SQL Server 2005?
More details:
I have some UDFs on Instance "VAXPServerA" in Database "DB1", for example "getSomething(number)".
It might look like this (for example):
----
CREATE FUNCTION [dbo].[getSomething] (
@theNumber int
)
RETURNS int AS
BEGIN
DECLARE @result int
SET @result = theNumber * 5
RETURN @result
END
----
I can call this function inside of the DB by typing:
----
SELECT dbo.getSomething(5)
----
I also can call the function from another DB on instance "VAXPServerA" by typing:
----
SELECT DB1.dbo.getSomething(5)
----
I have a second instance called "VAXPServerB", and "VAXPServerA" is a linked server in "VAXPServerB". I can do selects and stuff, it works fine.
But when I want to call the function...
----
SELECT [VAXPServerA].DB1.dbo.getSomething(5)
----
I obtain the next error:
Msg. 207, Level 16, State 1, Line 1
The column name 'VAXPServerA' is not valid.
Any hint?
Thanks in advance
David
I have a bunch of SPs that all rely on a UDF that parses a commadelimitted list of numbers into a table. Everything was working fine,but now my application is growing and I'm starting to approach the 8000character limit of the varChar variable used to store the list.I would like to change the UDF only and avoid having to dig through allof my stored procedures. I was hoping to use the text datatype toallow for much larger lists, but I am unable to perform anymanipulations necessary to parse the list into a table. I have triedPATINDEX, but it alone is not enough without the text maniuplations andI don't think the sp_xml_preparedocument can be used in a UDF.Anyone with any thoughts on managing large arrays in t-sql?thanks,Matt Weiner
View 2 Replies View RelatedHello Folks,I encountered a problem with SQL server 2000 and UDFs.I have a scalar UDF and a table UDF where I would like the scalar UDFto provide the argument for the table UDF like in:SELECT*FROMtransaction_tWHEREtrxn_gu_id in (select get_trxns_for_quarter(get_current_quarter( GetDate() ) ))'get_current_quarter' returns an integer which is a GUID in a tablecontaining business quarter definitions, like start date, end date.'get_current_quarter' is a scalar UDF.'get_trxns_for_quarter' will then get all transctions that fall intothat quarter and return their GUID's in a table.'get_trxns_for_quarter' is a table UDF.This doesn't seem to work at all. Regardless whether I provide thenamespace (schema) calling the scalar UDF or not. Error message isjust different.Both functions operate correctly invoked un-nested.The whole expression does work fine if I turn 'get_trxns_for_quarter'into a scalar UDF as well, e.g. by returning just one trxn_gu_id withe.g. MAX() in a scalar datatype. But of course that's no good to me.It also works fine if I select the result of 'get_current_quarter'into a variable and pass that variable into 'get_trxns_for_quarter'.But that's no good to me either since then I cannot use the wholething embedded into other SELECT clauses.Both UDF's are non-deterministic but I couldnt see how that would havean impact anyway.Never mind the syntax on that example or anyhting, I tried all theobvious and not so obvious stuff and it really seems to come down tothe fact that one UDF is scalar and the other one is not. However, Idid not come across any type of information saying that this cannot bedone.Have you any ideas?Any help would be greatly appreciated.Carsten
View 6 Replies View RelatedI am in the process of building my first "large scale" database system (after 15+ years of developing Windows Apps and Web Apps) - so I am very VERY "Green" when it comes to Database development & SQL et al.
A little context setting: I am building a multi-tier Statistical Analysis & Reporting system where the "end product" will be Reports created in Reporting Services. There are a ton of business rules that I am implementing in a Business Logic Tier (hidden from the "end user" by a Data Access Tier) comprised of SQL in the form of UDFs (scalar) and Views.
The question: I have been reading that UDFs cause a performance hit compared to things like in-line functions. Alot of the Rules (implemented as Scalar UDFs) build on each other so that the output of UDF #1 is used as input to UDF #2.
So far I am implementing the Business Logic as a hierarchy of Views (7 Views to be exact) with each view implementing multiple Rules; each Rule basically a Scalar UDF. Below is an example of what I am doing:
Example
View #1 -> Select A, B, C, funcX1(A) as ValueX1, funcY1(B, C) as ValueY1 FROM someView
Then
View #2 -> Select A, B, C, ValueX1, ValueY1, funcX2 (ValueX1) as ValueX2, funcY2(ValueY2) as ValueY2 FROM View#1
Currently I have a hierarchy of 7 views that each use UDFs to implement the Business Rules, where the value calculated from a UDF in one View is used as input to UDF in a View further down the Hierarchy.
Is there a better way of implementing all of the Rules instead of using multiple Views with a bunch of UDFs?
The "end product" dataset is then exposed as a Stored Procedure to the reports in Reporting Services.
Any help would be GREATLY appreciated.
Thanks!
- marty
There is something very strange going on here. Tested with ADO 2.7 andMSDE/2000. At first, things look quite sensible.You have a simple SQL query, let's sayselect * from mytab where col1 = 1234Now, let's write a simple VB program to do this query back to anMSDE/2000 database on our local machine. Effectively, we'llrs.open sSQLrs.closeand do that 1,000 times. We wont bother fetching the result set, itisn't important in this example.No problem. On my machine this takes around 1.6 seconds and modifyingthe code so that the column value in the where clause changes eachtime (i.e col1 = nnnn), doesn't make a substantial difference to thistime. Well, that all seems reasonable, so moving right along...Now we do it with a stored procedurecreate procedure proctest(@id int)asselect * from mytab where col1 = @idand we now find that executingproctest nnnn1,000 times takes around 1.6 seconds whether or not the argumentchanges. So far so good. No obvious saving, but then we wouldn'texpect any. The query is very simple, after all.Well, get to the point!Now create a table-returning UDFcreate function functest(@id int) returns table asreturn(select * from mytab where col1 = @id)try calling that 1,000 times asselect * from functest(nnnn)and we get around 5.5 seconds on my machine if the argument changes,otherwise 1.6 seconds if it remains the same for each call.Hmm, looks like the query plan is discarded if the argument changes.Well, that's fair enough I guess. UDFs might well be more expensive...gotta be careful about using them. It's odd that discarding the queryplan seems to be SO expensive, but hey, waddya expect?. (perhaps theUDF is completely rebuilt, who knows)last test, then. Create an SP that calls the UDFcreate procedure proctest1(@id int)asselect * from functest(@id)Ok, here's the $64,000 question. How long will this take if @idchanges each time. The raw UDF took 5.5 seconds, remember, so thisshould be slightly slower.But... IT IS NOT.. It takes 1.6 seconds whether or not @id changes.Somehow, the UDF becomes FOUR TIMES more efficient when wrapped in anSP.My theory, which I stress is not entirely scientific, goes somethinglike this:-I deduce that SQL Server decides to reuse the query plan in thiscircumstance but does NOT when the UDF is called directly. This iscounter-intuitive but it may be because SQL Server's query parser istuned for conventional SQL i.e it can saywell, I've gotselect * from mytab WHERE [something or other]and now I've gotselect * from mytab WHERE [something else]so I can probably re-use the query plan from last time. (I don't knowif it is this clever, but it does seem to know when twotextually-different queries have some degree of commonality)Whereas withselect * from UDF(arg1)andselect * from UDF(arg2)it goes... hmm, mebbe not.... I better not risk it.But withsp_something arg1andsp_something arg2it goes... yup, i'll just go call it... and because the SP was alreadycompiled, the internal call to the UDF already has a query plan.Anyway, that's the theory. For more complex UDFs, by the way, theperformance increase can be a lot more substantial. On a big complexUDF with a bunch of joins, I measured a tenfold increase inperformance just by wrapping it in an SP, as above.Obviously, wrapping a UDF in an SP isn't generally a good thing; theidea of UDFs is to allow the column list and where clause to filterthe rowset of the UDF, but if you are repeatedly calling the UDF withthe same where clause and column list, this will make it a *lot*faster.
View 3 Replies View Related