With SQL2005 SP2, we are seeing that when auto stats run on one or more indexes of a large table (1.5M rows), then immediately the stored proc using that table starts acting as if the query plan is no longer any good. This causes a drastic slowdown in response time and a corresponding increase of table reads to get the data. E.g, the next execution of the procedure after the auto stats kick in goes from 355 reads to 755000 reads (as depicted by Profiler). Generally, there are about 25 people using the DB at any one time. They connect through a mid-tier VB component.
I tried adding WITH RECOMPILE to the stored proc in question, but that caused almost all executions to run at the higher number. I thought that the WITH RECOMPILE hint would create a new query plan for each execution of the procedure and that plan would the the latest and greatest. Perhaps it did, but most users got stuck with the higher number of reads anyway. After taking the hint out, everyone went back to getting the 335 number and quick response times.
What we are wrestling with is that when those auto stats hit, it really messes up everyone until we manually recompile the procedure. Daily we delete all records in the table that are over 45 days old, so the table stays pretty much the same size. We also set the recompile flag to cause a new plan to be generated that will reflect the smaller amount of data. Should we also run a stats update before recompiling the procedure? Profiler has been very helpful in capturing what is going on, so I think I have a good handle on that. However, I don't understand why WITH RECOMPILE produced a messed up plan for everyone. The compile itself seems to take only 1 ms when done from the query screen.
I was wondering if anyone had an concrete information about if there is a problem with having too many stored procedures or plans in the cache? Obviously there is an impact on memory but if we can ignore that for the time being, does SQL perform just as well with 100 query plans as it does with 10's millions of plans?
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.
Hi all, I have an dtsx (SSIS) for "clone" manually Sql server database to another.
How I copy all stats from one database to another ? I have problem with "auto stats".
When I try DROP statitics for auto stats I get this error:
No se puede DROP el Ãndice 'dbo.ACTIVIDAD_PROVEEDOR.PK_ACTIVIDAD_PROVEEDOR'. No es una colección de estadÃsticas.
Cannot DROP index 'dbo.ACTIVIDAD_PROVEEDOR.PK_ACTIVIDAD_PROVEEDOR'. Not statitics collection.
What can I do ??
-- Get Stats list SELECT '[' + SCHEMA_NAME(tbl.schema_id) + '].[' + tbl.name + ']' AS [Table_Name_With_Schema], '[' + st.name + ']' AS [Name], '' + SCHEMA_NAME(tbl.schema_id) + '.' + tbl.name + '' + '.' + st.name + '' AS [Estadistica] FROM sys.tables AS tbl INNER JOIN sys.stats st ON st.object_id=tbl.object_id ORDER BY [Table_Name_With_Schema] ASC,[Name] ASC
Thanks in advance, any help will be appreciated, regards, greetings
Hi all, I have an dtsx (SSIS) for "clone" manually Sql server database to another.
How I copy all stats from one database to another ? I have problem with "auto stats".
When I try DROP statitics for auto stats I get this error:
No se puede DROP el Ãndice 'dbo.ACTIVIDAD_PROVEEDOR.PK_ACTIVIDAD_PROVEEDOR'. No es una colección de estadÃsticas.
Cannot DROP index 'dbo.ACTIVIDAD_PROVEEDOR.PK_ACTIVIDAD_PROVEEDOR'. Not statitics collection.
What can I do ??
-- Get Stats list SELECT '[' + SCHEMA_NAME(tbl.schema_id) + '].[' + tbl.name + ']' AS [Table_Name_With_Schema], '[' + st.name + ']' AS [Name], '' + SCHEMA_NAME(tbl.schema_id) + '.' + tbl.name + '' + '.' + st.name + '' AS [Estadistica] FROM sys.tables AS tbl INNER JOIN sys.stats st ON st.object_id=tbl.object_id ORDER BY [Table_Name_With_Schema] ASC,[Name] ASC
Thanks in advance, any help will be appreciated, regards, greetings
Does anyone know how to tell how long it took for an auto update statistics to run? I looked under DBCC Show_Statistics and it shows the time the stats were last updated, but not how long it took to update them. Thanks.
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON, FILLFACTOR = 90) ON [PRIMARY]
) ON [PRIMARY]
--Created an insert table DECLARE @COUNT INT
SET @COUNT = 1
WHILE @COUNT <= 1000
begin
insert into CUSTOMER (CUSTOMER_ID, CUSTOMER_NAME)
VALUES (@COUNT, '12345678901234567890')
SET @COUNT = @COUNT + 1
END
Look at Tables then statistics the statistics are empty so i fire update statistics and see 1000 rows in here.
I run again the insert script DECLARE @COUNT INT
SET @COUNT = 1001
WHILE @COUNT <= 2000
begin
insert into CUSTOMER (CUSTOMER_ID, CUSTOMER_NAME)
VALUES (@COUNT, '12345678901234567890')
SET @COUNT = @COUNT + 1
END
Look again at statistics it not firing.
If i do select * from CUSTOMER where CUSTOMER_ID = '2000' then go checks statictics it works.
I was under the impression that when you do insert or delete, update then the statistics are fired.
The sys.sysindexes rowmodctr shows the 1000 rows.
I checked the conditions that sql fires if the no of rows int able > 6 and < 500 then updates when 500 mods made. Also if row > 500 auto update done when 500 = 20% are added
So both are met.
Anyone other any other suggestions about the auto stats ?
Recently a production server suffered a critical blocking period and I wanted to know if I could solicit some input. It seems that a stored procedure was in the middle of recompiling while and auto update statistics started. This caused blocking for like an hour on the single object (stored procedure) that was originally called. The table that the update occurred on and that the stored procedure is reading form is quite large. It is 2 mil rows and about 140 columns wide. Some info from sysprocesses is below. The table alone takes up almost 4GB of space, when looking at sp_spaceused. I have some questions. 1. Can the update statistics for a '_WA%' stats cause blocking on a table? 2. Does an update stats on an index survive a restart of SQL server? We tried restarting, but the blocking did not end. 3. If the stored procedure is running under a compile, can the server automatically start an update stats and cause the stored procedure to wait? 4. Can the server automatically start an update stats on more than one column stats at a time, causing one to be blocked by the other? 5. We had never seen this issue before going to SQL2K clustering. Is this something specific to SQL2K and not SQL7 ?
Thanks for your input. John Lee
This is the lock info for the blocking processes.
spid dbid ObjId IndId Type Resource Mode Status name ------ ------ ----------- ------ ---- ---------------- -------- ------ ------------------------- 142 7 2 1 KEY (6f00035ef42b) S GRANT sysindexes 142 7 2 1 KEY (6f00035ef42b) S GRANT sysindexes 142 7 421576540 0 TAB Sch-S GRANT tJob 142 7 1141579105 0 TAB Sch-S GRANT tPatient_info 142 7 1141579105 0 TAB [UPD-STATS] Sch-M GRANT tPatient_info 142 7 1659921035 0 TAB [COMPILE] X GRANT iDBGetPatInfoRecord 142 7 1659921035 0 TAB Sch-S GRANT iDBGetPatInfoRecord
These are the processes that are being blocked:
spid ------ 137 140
Below this is a snapshot of all the SQL processes on the server being blocked. Save the report and send to the whole database group.
Does anyone know an easier way to create DB Maint Plans instead of having to create each one manually? i.e. Copy Maint Plans from 1 server to another automatically. Does anyone know of a quick export method.
Since upgrading from SQL Server Management Studio 2008 R2, I've noticed that it no longer autosaves queries that have not been manually saved first. If a file has been manually saved the autorecover files end up in the following directory:
%appdata%MicrosoftSQL Server Management Studio11.0AutoRecoverDatSolution1
However, I have ended up in the situation where I have unsaved queries when my computer has crashed and have not been able to recover them.
I have also found references to .sql files stored in temp files in the following directory, but the files here seem to be very haphazardly caught:
Hello all, I have a report with a table and a chart. It uses dataset1 as the data source. All works fine. I create a new dataset called dataset2. The queries are exactly the same. The only differences between the 2 datasets is the database server and the fact that one of the columns is a smallint (in dataset2) and an int(in Dataset1) I change the datasetName property of both the table and the chart to use dataset2. When I run the report I get a conversion error stating that there was an overflow of int2 while using dataset1. I have verified the report is not using dataset1 anywhere. If I delete dataset1 and run the report the error goes away. If I add it back, I get the error again. Why is the report looking at dataset1 if it is not referenced at all in the report? Does SQL RS cache the datasets and verify each when it compiles?
I am using SQL server 2005. I have a VIEW that joins several tables. One of the table's column can be added dynamically by the user from a GUI interface. However, after a column is added, it does not show up in the VIEW immediately. It will take a while (I haven't figured out exactly how long) before the extra column shows up as the execution result of the VIEW. So it seems like SQL server is caching that VIEW's schema. Is there anyway I can make this view always comes back with the latest schema? Thanks a lot! Penn
I want to check the performance of m query and i just want to remove cached query results. Is there any suggestion how can i do this. I just want to check after each modificatin how much improvement in performance
Phil, great links, really helpful and appreciated.
I just need to verify one thing on the lookup method: --One of the lookup methods people were discussing is non-cached lookup -- which seem to be evaluated to be the fastest. Is the non-cached the default of LookUp transformation? and when I wanted the lookup method to be cached, I need to go into the Advance tab and set it to however %, right? thanks.
I'm trying to understand the cases where it's more interesting to use snapshot and when it's more interesting to use cached instances.
If I have 100 users trying to reach a report, is it better to use snaphsot or cache instance ? In both case, the 100 users will have the same report result. And what about the performance, are they similar ?
I would like to know what is the difference between a snapshot and a cached instance in SSRS?
Which one has the best performances and which one is the best for multiple users and reports containing parameters (the parameters are then passed in the where clause of the sql code; ex: WHERE IN(@param1))?
Is it possible to keep a Cached Lookup in memory when executing multiple Data Flows? Executing DFT€™s in parallel will cache and use the same LOOKUP statement. But what if I€™m executing the DFT sequentially, can I keep the LOOKUP from the first DFT in memory for the second DFT? For example, in my case, I€™m caching a lookup against the Customer dimension for invoices. The second DFT then processes credits and again does a lookup against the Customer dimension. I want to use the cached Customer records from the first DFT.
Parameterized queries are only allowed on partial or none cache style lookup transforms, not 'full' ones. Is there some "trick" to parameterizing a full cache lookup, or should the join simply be done at the source, obviating the need for a full cache lookup at all (other suggestion certainly welcome)
More particularly, I'd like to use the lookup transform in a surrogate key pipeline. However, the dimension is large (900 million rows), so its would be useful to restrict the lookup transform's cache by a join to the source.
For example:
Source query is: select a,b,c from t where z=@filter (20,000 rows)
Lookup transform query: select surrogate_key,business_key from dimension (900 M rows, not tenable)
I have an MS SQL Server table with a Job Number field I need this field to start at a certain number then auto increment from there. Is there a way to do this programatically or within MSDE?
I have a procedure that generates dynamic sql and then executes via the execute(strSQL) syntax. BOL states that if I use sp_executesql with hard-typed parameters passed in variables, the query optimizer will 'probably' match the sql statement with the cached execution path, thus avoiding recompilation and speeding up the results for heavily run procedures.
Can anyone tell me if this is also true if the sql references an object on a linked sql server 2000 database? Technically, the sql is exactly the same, but I'm unsure if there is some exception due to the way linked objects are processed.
Hello, This is my first post, and I'm hoping you all can help.
Using Reporting Services 2005, I have several reports that use embedded images. All images render fine when the report execution is set to: 1) Always run this report with the most recent data 1a) Do not cache temporary copies of this report
However, when I change the execution to either a Cache or a Snapshot, the images and some charts render as red "X" placeholders. This is sometimes remedied when the user clicks the page refresh button, but not always.
Of course, I could just have all the concurrent users use the uncached report that hits the OLAP server, but that would be highly inefficient, and just plain slow.
I am frequently going back and forth between the making changes to my reports and previewing those changes, all from within visual studio without publishing the reports each time.
The problem is that frequently changes that I make are not reflected in the preview window unless I either close out of visual studio completely or I wait some length of time, (I'm not sure how long exactly, but 1/2 an hour seems to always do the trick.)
Is there a way to clear any cache and force visual studio to completely reprocess a report?
At least when it is formatting changes I can identify whether the change has stuck, but when I'm fixing bugs in the code, I can't tell if I didn't fix it or if the change just hasn't taken effect.
I'm working on a reporting tool that could bring back hundreds of thousands of results back at once. I need some way to run the actual query only once a day, and then the reporting tool would just pull back this cached results. To be short, I need to figure out how to do this using a minimum amount of resources. Would a DataView work with something like this? How would I have it update only once a day? I appreciate any advice!
I have a datagridview bound to a table that is part of an Entity Framework model. A user can edit data in the datagridview and save the changes back to SQL. But, there is a stored procedure that can also change the data, in SQL, not in the datagridview. When I try to "refresh" the datagridview the linq query always returned the older cached data. Here's the code that I have tried using to force EF to pull retrieve new data:
// now refresh the maintenance datagridview data source using (var context = new spdwEntities()) { var maintData = from o in spdwContext.MR_EquipmentCheck where o.ProdDate == editDate orderby o.Caster, o.Strand select o; mnt_DGV.DataSource = maintData; }
When I debug, I can see that the SQL table has the updated data in it, but when this snippet of code runs, maintData has the old data in it.
every morning I have 7-10 identical messages in error log
1. Configuration option 'allow updates' changed from 1 to 0. Run the RECONFIGURE statement to install.. 2. Error: 15457, Severity: 0, State: 1 3. Configuration option 'allow updates' changed from 0 to 1. Run the RECONFIGURE statement to install..
It is standby server with custom log shipping and DTS transfering logins every 15 min.
We think we're having performance problems, and among the areas of investigations is the tempdb database. Since it resets itself after SQL is restarted, is there a way to find out how big it has grown in the past ? Does leaving it at the default size cause a performace hit ?? Right now it's 8.75 MB, with 7.38 MB available, which sounds pretty harmless.
Everywhere I read, it states that running SQL Profiler can affect performance of your SQL Server. My question is - how much of an impact will it really make? Will I see a 1% degredation in peformance? 5%? 50%? I haven't been able to find a good answer. We currently have SQL Profiler running all day long for almost 3 years, and the databases are still humming.
Is it the amount of data you are requesting from the trace that affects performance? There are some compliance tools out there (Idera Compliance Manager, IPLocks, etc) that run a profiler trace to get data. There are other DBAs in my organization who don't want to use them because "profiler traces will degrade my SQL Server performance". How true is this really.
Any help I can get would be extremely appreciated.
Does multiplication with 1 affect query performance?I have a a stored procedure that converts results to another unit if required. In alternative 1 below, the results are returned with a separate select statement if no conversion is necessary - in other words, no multiplication with a conversion factor is required. However, the code is not very nice since I need to repeat the select statement again in case a conversion is required, this time including the conversion factor.Alternative 2 uses cleaner-looking code. The conversion factor is set to 1 if no conversion is required, and a single SELECT statement is used to return the data. The @factor variable is defined as a float.I would rather use alternative 2, but I wonder if there is any performance penalty for doing that if no conversion is required since the results are always multiplied with the @factor? Or can SQL server somehow understand that @factor = 1 and no multiplication is required?--- Alternative 1: ---IF @fromunit_sid = @tounit_sid-- Return unconverted results SELECT ISNULL(ls_totalWaterConsumption,0) AS ls_totalWaterConsumption,ls_theoreticalWaterConsumption AS ls_theoretical_WaterConsumption,ls_totalWaterConsumption - ls_theoreticalWaterConsumption AS ls_extra_WaterConsumption FROM Results WHERE scenario_id = @scenario_idELSEBEGIN -- Get conversion factor EXEC getConversionFactor @fromunit_sid, @tounit_sid, @factor OUTPUT -- Get the converted results SELECT ISNULL(ls_totalWaterConsumption * @factor,0) AS ls_totalWaterConsumption, ls_theoreticalWaterConsumption * @factor AS ls_theoretical_WaterConsumption, (ls_totalWaterConsumption - ls_theoreticalWaterConsumption) * @factor AS ls_extra_WaterConsumptionFROM Results WHERE scenario_id = @scenario_idEND --- Alternative 2: ---IF @fromunit_sid = @tounit_sidSET @factor = 1ELSE -- Get conversion factor EXEC getConversionFactor @fromunit_sid, @tounit_sid, @factor OUTPUT
-- Get the converted results SELECT ISNULL(ls_totalWaterConsumption * @factor,0) AS ls_totalWaterConsumption, ls_theoreticalWaterConsumption * @factor AS ls_theoretical_WaterConsumption, (ls_totalWaterConsumption - ls_theoreticalWaterConsumption) * @factor AS ls_extra_WaterConsumptionFROM Results WHERE scenario_id = @scenario_id And another question: is using an IF function considerably faster than making a call to another stored procedure?In alternative 2 above I use an IF statement to check if @fromunit_sid = @tounit_sid, and . But in fact the function getConversionFactor that I'm calling does exactly the same thing: if I pass in identical from- and to-values, it simply returns 1, so I could omit the IF statement completely and just use alternative 3. But is it slower?--- Alternative 3 -- Get conversion factor EXEC getConversionFactor @fromunit_sid, @tounit_sid, @factor OUTPUT
Hi, I am wrote the following code in one store procedure called p_bcp_all. and then scheduled it to run over night. what if the first two bcp were successful but the third one failed. Is the whole procedure going to fail? also what if the first one failed, is the rest of the code going to be executed have the bcp process going for the second and the third table? thanks for your adivce
regards Ali
create procedure p_bcp_all as Exec master..xp_cmdshell "bcp servername..tblone in d:blone.txt /fd:formatfileblone.fmt /servername /Usa /password/b250000 /a8000"