We are trying to solve a real puzzle. We have a stored procedure that
exhibits *drastically* different execution times depending on how its
executed.
When run from QA, it can take as little as 3 seconds. When it is
called from an Excel vba application, it can take up to 180 seconds.
Although, at other times, it can take as little as 20 seconds from
Excel.
Here's a little background. The 180 second response time *usually*
occurs after a data load into a table that is referenced by the stored
procedure.
A check of DBCC show_statistics shows that the statistics DO get
updated after a large amount of data is loaded into the table.
*** So, my first question is, does the updated statistics force a
recompile of the stored procedure?
Next, we checked syscacheobjects to see what was going on with the
execution plan for this stored procedure. What I expected to see was
ONE execution plan for the stored procedure.
This is not the case at all. What is happening is that TWO separate
COMPILED PLANs are being created, depending on whether the sp is run
from QA or from Excel.
In addition, there are several EXECUTABLE PLANs that correspond to the
two COMPILED PLANs. Depending on *where* the sp is run, the usecount
increases for the various EXECUTABLE PLANS.
To me, this does not make any sense! Why are there *multiple* compile
and executable plans for the SAME sp?
One theory we have is, that we need to call the sp with the dbo
qualifier, ie) EXEC dbo.sp
Has anyone seen this? I just want to get to the bottom of this and
find out why sometimes the query takes 180 seconds and other times
only takes 3 seconds!!
I think not. Microsoft says it is possible: one for parallel and one for serial execution. Don't believe that's possible for a stored procedure to change execution plans on the fly. Have an on-going problem with timeout occurring with an application and narrowed the culprit to a stored procedure. I couldn't find any obvious issues database wise, no locks, etc. so I recompiled (altered) the sproc without making any changes and the issue cleared for a couple days.
It happened again to day, and so I recompiled (altered) the sproc and it went away again. No code changes to both application (so they say) and stored procedure. I ran the below code snippet to check for sprocs with multiple cached plans and the offending one came up on a short list. So, my question is, Is it one sproc per query plan or can there be more than one. I understand the connection issues.
Hi I am slowly getting to grips with SQL Server. As a part of this, I have been attempting to work on producing more efficient queries. This post is regarding what appears to be a discrepancy between the SQL Server execution plan and the actual time taken by a query to run. My brief is to produce an attendance system for an education establishment (I presume you know I'm not an A-Level student completing a project :p ). Circa 1.5m rows per annum, testing with ~3m rows currently. College_Year could strictly be inferred from the AttDateTime however it is included as a field because it a part of just about every PK this table is ever likely to be linked to. Indexes are not fully optimised yet. Table:CREATE TABLE [dbo].[AttendanceDets] ([College_Year] [smallint] NOT NULL ,[Group_Code] [char] (12) COLLATE SQL_Latin1_General_CP1_CI_AS NOT NULL ,[Student_ID] [char] (8) COLLATE SQL_Latin1_General_CP1_CI_AS NOT NULL ,[Session_Date] [datetime] NOT NULL ,[Start_Time] [datetime] NOT NULL ,[Att_Code] [char] (1) COLLATE SQL_Latin1_General_CP1_CI_AS NOT NULL ) ON [PRIMARY]GO CREATE CLUSTERED INDEX [IX_AltPK_Clust_AttendanceDets] ON [dbo].[AttendanceDets]([College_Year], [Group_Code], [Student_ID], [Session_Date], [Att_Code]) ON [PRIMARY]GO CREATE INDEX [All] ON [dbo].[AttendanceDets]([College_Year], [Group_Code], [Student_ID], [Session_Date], [Start_Time], [Att_Code]) ON [PRIMARY]GO CREATE INDEX [IX_AttendanceDets] ON [dbo].[AttendanceDets]([Att_Code]) ON [PRIMARY]GOALL inserts are via an overnight sproc - data comes from a third party system. Group_Code is 12 chars (no more no less), student_ID 8 chars (no more no less). I have created a simple sproc. I am using this as a benchmark against which I am testing my options. I appreciate that this sproc is an inefficient jack of all trades - it has been designed as such so I can compare its performance to more specific sprocs and possibly some dynamic SQL. Sproc:CREATE PROCEDURE [dbo].[CAMsp_Att] @College_Year AS SmallInt,@Student_ID AS VarChar(8) = '________', @Group_Code AS VarChar(12) = '____________', @Start_Date AS DateTime = '1950/01/01', @End_Date as DateTime = '2020/01/01', @Att_Code AS VarChar(1) = '_' AS IF @Start_Date = '1950/01/01'SET @Start_Date = CAST(CAST(@College_Year AS Char(4)) + '/08/31' AS DateTime) IF @End_Date = '2020/01/01'SET @End_Date = CAST(CAST(@College_Year +1 AS Char(4)) + '/07/31' AS DateTime) SELECT College_Year, Group_Code, Student_ID, Session_Date, Start_Time, Att_Code FROM dbo.AttendanceDets WHERE College_Year = @College_YearAND Group_Code LIKE @Group_CodeAND Student_ID LIKE @Student_IDAND Session_Date <= @End_DateAND Session_Date >=@Start_DateAND Att_Code LIKE @Att_CodeGOMy confusion lies with running the below script with Show Execution Plan:--SET SHOWPLAN_TEXT ON--Go DECLARE @Time as DateTime Set @Time = GetDate() select College_Year, group_code, Student_ID, Session_Date, Start_Time, Att_Code from attendanceDetswhere College_Year = 2005 AND group_code LIKE '____________' AND Student_ID LIKE '________'AND Session_Date <= '2005-11-16' AND Session_Date >= '2005-11-16' AND Att_Code LIKE '_' Print 'First query took: ' + CAST(DATEDIFF(ms, @Time, GETDATE()) AS VarCHar(5)) + ' milli-Seconds' Set @Time = GetDate() EXEC CAMsp_Att @College_Year = 2005, @Start_Date = '2005-11-16', @End_Date = '2005-11-16' Print 'Second query took: ' + CAST(DATEDIFF(ms, @Time, GETDATE()) AS VarCHar(5)) + ' milli-Seconds'GO --SET SHOWPLAN_TEXT OFF--GOThe execution plan for the first query appears miles more costly than the sproc yet it is effectively the same query with no parameters. However, my understanding is the cached plan substitutes literals for parameters anyway. In any case - the first query cost is listed as 99.52% of the batch, the sproc 0.48% (comparing the IO, cpu costs etc support this). BUT the text output is:(10639 row(s) affected) First query took: 596 milli-Seconds (10639 row(s) affected) Second query took: 2856 milli-SecondsI appreciate that logical and physical performance are not one and the same but can why is there such a huge discrepancy between the two? They are tested on a dedicated test server, and repeated running and switching the order of the queries elicits the same results. Sample data can be provided if requested but I assumed it would not shed much light. BTW - I know that additional indexes can bring the plans and execution time closer together - my question is more about the concept. If you've made it this far - many thanks.If you can enlighten me - infinite thanks.
I have two schematically identical databases on the same MS SQL 2000 server. The differences in the data are very slight. Here is my problem: the identical query has totally different execution plans on the different databases. One is (in my opinion) correct, the other causes the query to take 60 times as long. This is not an exaggeration, on the quick DB the query takes 3 seconds, on the other DB it takes 3 minutes. I have tried the following to help the optimizer pick a better execution plan on the slow db:
rebuild the indexes dbcc indexdefrag update statistics
I CAN put in a hint to cause the query to execute faster, but my employer now knows about the problem and he (and I) want to know WHY this is happening.
HiCan you give me sone pointers to where I can get more information aboutthe various operations like index seeks,Bookmark Lookups,ClusteredIndex Scan in an execution plan.ThanksRagu
I'm looking for assistance on a problem with SQL Server. We have adatabase where a particular query returns about 3000 rows. This querytakes about 2 minutes on most machines, which is fine in thissituation. But on another machine (just one machine), it can run forover 30 minutes and not return. I ran it in Query Analyzer and it wasreturning about 70 rows every 45-90 seconds, which is completelyunacceptable.(I'm a developer, not a DBA, so bear with me here.)I ran an estimated execution plan for this database on each machine,and the "good" one contains lots of parallelism stuff, in particularthe third box in from the left. The "bad" one contains a "Nested Loop"at that position, and NO parallelism.We don't know exactly when this started happening, but we DO know thatsome security updates have been installed on this machine (it's at theclient location), and also SP1 for Office 2003.So it looks like parallelism has been turned off by one of these fixes.Where do we look for how to turn it back on? This is on SQL Server2000 SP3.Thanks for any help you might have for me!Christine Wolak -- SPL WorldGroup --Join Bytes!
I have a table TableA with few million rows. When I query TableA , the execution plans changes based on the input parameter as shown below . Why this happens ? How to resolve this ? Any inputs would be appreciated.
SELECT * FROM TableA WHERE Column1 = 1 => SELECT -> Clustered Index Scan (100%)
SELECT * FROM TableA WHERE Column1 = 2 => SELECT -> Clustered Index Scan (100%)
SELECT * FROM TableA WHERE Column1 = 3 => SELECT -> Parallelism (3%) -> Clustered Index Scan (97%)
SELECT * FROM TableA WHERE Column1 = 4 => SELECT -> Nested Loops -> Index Seek (50%) -> Clustered Index Seek (50%) (takes a very long time to retrieve the records)
Pls tell me where i will be able to find a good material on interpreting the Execution plans................how do i compare 2 diff plans for Quries written in 2 diff ways...giving same output
In using ADO to connect to SQL Server, I'm trying to retrieve multiple datasets AND statistics that are usually returned via the OnInfoMessage event. For those that are familiar with SQL Server, I need the results returned by the SET STATISTICS IO ON and SET STATISTICS PROFILE ON options. Anyone had any luck doing this before?
I've been working with SQL Server 2005 for a while now and I've noticed some odd behavior that I want to bounce of other members of the community. I should preface that I've been a forum viewer (and occasional contributer) here at SQL Team for a while and I've naturally developed a keen sense for optimizations.
Fundamentally, longer stored procedures with perfectly fine/optimized execution plans are inconsistent with real world performance. In some of these cases, a low subtree cost on a 4 core machine with 16gb of ram and 2 15 drive SAS arrays with little load takes excessively long to run or in some cases doesn't complete.
This isn't due to blocking or resource bottlenecks as I'm quite familiar with built in tools to troubleshoot and resolve those issues. In all cases, I am able to rearchitect the stored procedure into a higher subtree cost variant and get reasonable performance, but it's frustrating to have to redo work and there seems to be no common theme other than longer multi-statement procedures.
I've used SQL Server 2000 extensively and did not notice this level of inconsistency in performance with that product version. Just wondering if others in the community have experiences similar or if I'm just crazy.
Hi there - hoping someone can help me here!I have a database that has been underperforming on a number of queriesrecently - in a test environment they take only a few seconds, but onthe live data they take up to a minute or so to run. This is using thesame data.Every evening a copy of the live data is copied to a backup 'snapshot'database on the same server and also, on this copy the queries onlytake a second or so to run. (This is testing through the QueryAnalyser)I've studied the execution plans for the same query on the snapshot dband the live db and they seem to be significantly different - why isthis? it's looking at the same data and exactly the same code!!Anybody got any ideas???
I'm trying to get an application finished that works like Query Analizer in terms of returning query plans and statistics.
Problem the co-author is having:
>In using ADO to connect to SQL Server, I'm trying to retrieve multiple >datasets AND statistics that are usually returned via the OnInfoMessage >event. For those that are familiar with SQL Server, I need the results >returned by the SET STATISTICS IO ON and SET STATISTICS PROFILE ON options. >Anyone had any luck doing this before?
Can anyone shed any light on this please?
Thanks.
BTW if anyone wants to take a look at the tool so far - to see what I'm delving into: http://81.130.213.94/myforum/forum_posts.asp?TID=78&PN=1
We know that a query execution plan exists for Stored Procedures in the Procedure Cache.
What about Views? Does a view have a query execution plan? We know that a View is a virtual table and that virtual table is populated when the view is induced but does it have a Query Execution Plan?
I have tried to find this info for Views in BOL but I cannot see it anywher in BOL.
SELECT * FROM a LEFT OUTER JOIN b ON a.id = b.id instead of
SELECT * FROM a LEFT JOIN b ON a.id = b.id
generates a different execution plan?
My query is more complex, but when I change "LEFT OUTER JOIN" to "LEFT JOIN" I get a different execution plan, which is absolutely baffling me! Especially considering everything I know and was able to research essentially said the "OUTER" is implied in "LEFT JOIN".
I have a stored procedure that suddenly started performing horribly.The query plan didn't look right to me, so I copy/pasted the code andran it (it's a single SELECT statement). That ran pretty well and useda query plan that made sense. Now, I know what you're all thinking...stored procedures have to optimize for variable parameters, etc.Here's what I've tried to fix the issue:1. Recompiled the stored procedure2. Created a new, but identical stored procedure3. Created the stored procedure with the RECOMPILE option4. Created the stored procedure with a hard-coded value instead ofaparameter5. Changed the stored procedure to use dynamic SQLIn every case, performance did not improve and the query plan remainedthe same (I could not easily confirm this with the dynamic SQLversion, but performance was still horrible).I am currently running UPDATE STATISTICS on all of the involvedtables, but that will take awhile.Any ideas?Thanks!-Tom.
Any way to invalidate cached query plans? I would rather target a specific query instead of invalidating all of them. Also any sql server setting that will cause cached query plans to invalidate even though only one character in the queries has changed?
exec sp_executesql N'select cast(5 as int) as DisplaySequence, mt.Description + '' '' + ct.Description as Source, c.FirstName + '' '' + c.LastName as Name, cus.CustomerNumber Code, c.companyname as "Company Name", a.Address1, a.Address2,
[code]....
In this query we have seen (on some databases) simply changing ‘@CustomerId int',@CustomerId=1065’ too ‘@customerId int',@customerId=1065’ fixed the a speed problem….just changed the case on the Customer bind parameter. On other servers this has no effect.the server is using an old cached query plan, but don’t know for sure.
way to invalidate cached query plans? I would rather target a specific query instead of invalidating all of them.
Also do you know of any sql server setting that will cause cached query plans to invalidate even though only one character in the queries has changed?
exec sp_executesql N'select cast(5 as int) as DisplaySequence, mt.Description + '' '' + ct.Description as Source,
[Code].....
In this query we have seen (on some databases) simply changing ‘@CustomerId int',@CustomerId=1065’ too ‘@customerId int',@customerId=1065’ fixed the a speed problem….just changed the case on the Customer bind parameter. On other servers this has no effect. I’m thinking the server is using an old cached query plan, but don’t know for sure.
Try this script to see what queries are taking over a second.To get some real output, you need a long-running query. Here's one(estimated to take over an hour):PRINT GETDATE()select count_big(*)from sys.objects s1, sys.objects s2, sys.objects s3,sys.objects s4, sys.objects s5PRINT GETDATE()Output is:session_id elapsed task_alloc task_dealloc runningSqlText FullSqlTextquery_plan51 32847 0 0 select count_big(*) from sys.objects s1, sys.objects s2,sys.objects s3, sys.objects s4, sys.objects s5 SQL PlanClicking on SQL opens the full SQL batch as a .txt file, including the PRINTstatementsClicking on Plan allows you to see the .sqlplan file in MSSMS========Title: Using a VB Script to show long-running queries, complete with queryplans.Today (July 14th), I found a query running for hours on a development box.Rather than kill it, I decided to use this opportunity to develop a scriptto show long-running queries, so I could see what was going on. (ReferenceRoy Carlson's article for the idea.)This script generates a web page which shows long-running queries with thecurrently-executing SQL command, full SQL text, and .sqlplan files. The fullSQL query text and the sqlplan file are output to files in your tempdirectory. If you have SQL Management Studio installed on the localcomputer, you should be able to open the .sqlplan to see the query plan ofthe whole batch for any statement.'LongestRunningQueries.vbs'By Aaron W. West, 7/14/2006'Idea from:'http://www.sqlservercentral.com/columnists/rcarlson/scriptedserversnapshot.asp'Reference: Troubleshooting Performance Problems in SQL Server 2005'http://www.microsoft.com/technet/prodtechnol/sql/2005/tsprfprb.mspxSub Main()Const MinimumMilliseconds = 1000Dim srvnameIf WScript.Arguments.count 0 Thensrvname = WScript.Arguments(0)Elsesrvname = InputBox ( "Enter the server Name", "Server", ".", VbOk)If srvname = "" ThenMsgBox("Cancelled")Exit SubEnd IfEnd IfConst adOpenStatic = 3Const adLockOptimistic = 3Dim i' making the connection to your sql server' change yourservername to match your serverSet conn = CreateObject("ADODB.Connection")Set rs = CreateObject("ADODB.Recordset")' this is using the trusted connection if you use sql logins' add username and password, but I would then encrypt this' using Windows Script Encoderconn.Open "Provider=SQLOLEDB;Data Source=" & _srvname & ";Trusted_Connection=Yes;Initial Catalog=Master;"' The query goes heresql = "select " & vbCrLf & _" t1.session_id, " & vbCrLf & _" t2.total_elapsed_time AS elapsed, " & vbCrLf & _" -- t1.request_id, " & vbCrLf & _" t1.task_alloc, " & vbCrLf & _" t1.task_dealloc, " & vbCrLf & _" -- t2.sql_handle, " & vbCrLf & _" -- t2.statement_start_offset, " & vbCrLf & _" -- t2.statement_end_offset, " & vbCrLf & _" -- t2.plan_handle," & vbCrLf & _" substring(sql.text, statement_start_offset/2, " & vbCrLf & _" CASE WHEN statement_end_offset<1 THEN 8000 " & vbCrLf & _" ELSE (statement_end_offset-statement_start_offset)/2 " & vbCrLf & _" END) AS runningSqlText," & vbCrLf & _" sql.text as FullSqlText," & vbCrLf & _" p.query_plan " & vbCrLf & _"from (Select session_id, " & vbCrLf & _" request_id, " & vbCrLf & _" sum(internal_objects_alloc_page_count) as task_alloc, " &vbCrLf & _" sum (internal_objects_dealloc_page_count) as task_dealloc " &vbCrLf & _" from sys.dm_db_task_space_usage " & vbCrLf & _" group by session_id, request_id) as t1, " & vbCrLf & _" sys.dm_exec_requests as t2 " & vbCrLf & _"cross apply sys.dm_exec_sql_text(t2.sql_handle) AS sql " & vbCrLf & _"cross apply sys.dm_exec_query_plan(t2.plan_handle) AS p " & vbCrLf & _"where t1.session_id = t2.session_id and " & vbCrLf & _" (t1.request_id = t2.request_id) " & vbCrLf & _" AND total_elapsed_time " & MinimumMilliseconds & vbCrLf & _"order by t1.task_alloc DESC"rs.Open sql, conn, adOpenStatic, adLockOptimistic'rs.MoveFirstpg = "<html><head><title>Top consuming queries</title></head>" & vbCrLfpg = pg & "<table border=1>" & vbCrLfIf Not rs.EOF Thenpg = pg & "<tr>"For Each col In rs.Fieldspg = pg & "<th>" & col.Name & "</th>"c = c + 1Nextpg = pg & "</tr>"Elsepg = pg & "Query returned no results"End Ifcols = cdim filenamedim WshShellset WshShell = WScript.CreateObject("WScript.Shell")Set WshSysEnv = WshShell.Environment("PROCESS")temp = WshShell.ExpandEnvironmentStrings(WshSysEnv("TEMP")) & ""filename = temp & filenameDim fso, fSet fso = CreateObject("Scripting.FileSystemObject")i = 0Dim cDo Until rs.EOFi = i + 1pg = pg & "<tr>"For c = 0 to cols-3pg = pg & "<td>" & RTrim(rs(c)) & "</td>"Next'Output FullSQL and Plan Text to files, provide links to themfilename = "topplan-sql" & i & ".txt"Set f = fso.CreateTextFile(temp & filename, True, True)f.Write rs(cols-2)f.Closepg = pg & "<td><a href=""" & filename & """>SQL</a>"filename = "topplan" & i & ".sqlplan"Set f = fso.CreateTextFile(temp & filename, True, True)f.Write rs(cols-1)f.Closepg = pg & "<td><a href=""" & filename & """>Plan</a>"'We could open them immediately, eg:'WshShell.run temp & filenamers.MoveNextpg = pg & "</tr>"Looppg = pg & "</table>"filename = temp & "topplans.htm"Set f = fso.CreateTextFile(filename, True, True)f.Write pgf.CloseDim oIESET oIE = CreateObject("InternetExplorer.Application")oIE.Visible = TrueoIE.Navigate(filename)'Alternate method:'WshShell.run filename' cleaning uprs.Closeconn.CloseSet WshShell = NothingSet oIE = NothingSet f = NothingEnd SubMain
Can I roll back certain query(insert/update) execution in one page if query (insert/update) in other page execution fails in asp.net.( I am using sqlserver 2000 as back end) scenario In a webpage1, I have insert query into master table and Page2 I have insert query to store data in sub table. I need to rollback the insert command execution for sub table ,if insert command to master table in web page1 is failed. (Query in webpage2 executes first, then only the query in webpage1) Can I use System. Transaction to solve this? Thanks in advance
Hi all I'm using Sql server 2000 and sometimes i need to run my Queries in Query analizer before using them in my application just to test them...BUT most of the time when i run a query in query-analizer for second time ,query analizer populates the result (records) more quicker then the first time. Apparently it caches the query !!! i don't know but for some reasons i dont't want this , so how can i prevent Query-analizer from doing so? Thanks in advance. Regards.
I compared view query plan with query plan if I run the same statementfrom view definition and get different results. View plan is moreexpensive and runs longer. View contains 4 inner joins, statisticsupdated for all tables. Any ideas?
SELECT *,isnull(dbo.fn_1(a),'') f_a, isnull(dbo.1(b),'') f_b FROM tablea with(nolock) WHERE ID = 12345 and ID<10000000 and dbo.fn_1(7)='asdfasdf' and Active='Y'
Does it effect performance as i am using a udf here? Also let me know the order in which the conditions will be applied.
I want to know is there any way by which we can come to know which query is going to be executed by the Engine prior to its execution. I think if this is possible then this will affect the performance. But It will be very helpful for me.
I have sql query to search for fields in a rather big view. If I execute the query in sql server enterprise manager, the results will be displayed in less than 6 seconds. However, if I execute it using asp.net, it will take very long (more than 2 minutes).
The query is a simple one like "SELECT * FROM myview WHERE name LIKE '%Microsoft%'". And the code I use to execute it in asp.net is
Dim dsRtn As DataSet Dim objConnection As OleDbConnection Try objConnection = GetOleDbConnection() objConnection.Open() Dim objDataAdapter As New OleDbDataAdapter(strSearch, objConnection) Dim objDataSet As New DataSet() objDataAdapter.Fill(objDataSet, strTableName) dsRtn = objDataSet Catch ex As Exception dsRtn = Nothing Finally If objConnection.State = ConnectionState.Open Then objConnection.Close() End If End Try
Where strSearch is the sql search string.
I don't have any problem using such code for other queries.
Could somebody suggest the cause of the problem and how to solve it? Thanks!
I am having a query where I am connecting to eight different tables using joins. When I join one table to another the speed of the execution becomes less. Even on my local server it is taking nearly 2 to 3 minutes to execute the query. How can I increase the speed of execution of my query.
I wanted to know whether we have an execution plan enabled in SQL 6.5 as we have it in SQL 7.0 and SQL 2000 . I.e when we execute a query and if we enable ' show execution plan 'then it creates a map and shows the vital statistics . If that is available on SQL 6.5 then i am missing that tool .
How can i have it installed on my SQL 6.5 server ??
