AppDev: Something You Should Know by Irena Kennedy : SQL

SQL Server Connection Leak Simulation Tool

irenak — Wed, 16 Sep 2009 23:41:00 GMT

Sometimes it’s necessary to test application performance in low available resource situations… I’ve seen tools that simulate low available memory, do network throttling, etc., but I have not come across one that simulates a lot of SQL Server connections...(read more)

SYSK 360: What’s taking up the space in my database? Also, a word about SQL reports…

irenak — Wed, 08 Aug 2007 03:37:00 GMT

So, you’ve updated usage stats on your database by running EXEC sp_spaceused @updateusage = N 'TRUE' But, you still can’t figure out why it’s reporting much more used space then the data that appears in your tables… Try running the following query to...(read more)

SYSK 359: How to Purge Workflow Tracking Data Inserted By SqlTrackingService

irenak — Mon, 06 Aug 2007 03:06:00 GMT

To my great surprise, after doing a search for a script that would delete the events logged by the almighty SqlTrackingService service, I didn’t find one… So, I quickly created it (see below), and it seems to do the job just fine given the current table...(read more)

SYSK 357: SQL Server Login Problem on Fresh Installation in Vista

irenak — Sun, 05 Aug 2007 19:41:00 GMT

I just installed SQL Server 2008 CTP (Katmai) and all the patches on my Vista laptop… However, trying to login using SQL Management Studio or osql using integrated security (and, yes, I’m an admin on my box) results in the “login failed” error. To resolve...(read more)

SYSK 349: What’s New in SQL 2008 (code named Katmai)?

irenak — Mon, 18 Jun 2007 15:52:00 GMT

Here are some of the new and improved features (extracted from David Campbell’s presentation) that will be available in SQL 2008:

- Transparent data encryption

- External key management

- Data auditing

- Pluggable CPU

- Transparent failover for data mirroring

- Declarative management framework

- Server group management

- Streamlined installation

- Enterprise system management

- Performance data collection

- System analysis

- Data compression

- Query optimization modes

- Resource governor

- Entity data model

- LINQ

- Visual entity designer

- Entity aware adapters

- SQL Server change tracking

- Synchronized programming model

- Visual studio support

- SQL Server conflict detection

- FILESTREAM data type

- Integrated full text search

- Sparse columns

- Large (> 1 page) user defined types

- Data/time data type

- LOCATION data type

- SPACIAL data type

- Virtual Earth integration

- Partitioned table parallelism

- Query optimizations

- Persistent lookups

- Change data capture

- Backup compression

- MERGE SQL statement

- Data profiling

- Star join

- Enterprise reporting engine

- Internet report deployment

- Block computations

- Scale out analysis

- BI platform management

- Export to Word and Excel

- Author reports in Word and Excel

- Report Builder enhancements

- TABLIX

- Rich formatted data

- Personalized perspectives

- … and many more

You can hear David’s webcast, and learning more about what those feature actually bring, visit http://www.microsoft.com/events/EventDetails.aspx?CMTYSvcSource=MSCOMMedia&Params=%7eCMTYDataSvcParams%5e%7earg+Name%3d%22ID%22+Value%3d%221032341072%22%2f%5e%7earg+Name%3d%22ProviderID%22+Value%3d%22A6B43178-497C-4225-BA42-DF595171F04C%22%2f%5e%7earg+Name%3d%22lang%22+Value%3d%22en%22%2f%5e%7earg+Name%3d%22cr%22+Value%3d%22US%22%2f%5e%7esParams%5e%7e%2fsParams%5e%7e%2fCMTYDataSvcParams%5e

or search by event id 1032341072 at http://msevents.microsoft.com/cui/AdvancedSearch.aspx?culture=en-US

If you’re short on time, you may choose to skip the first 18 minutes containing introduction, a bit of data processing history and the new scenario-based rather than feature-based product development.

Here are some main time points:

00:00 – 18:00

Intro, history

18:00 – 23:30

New & improved features related to the secure, trusted platform, policy-based management, and optimized and predictable performance scenarios

23:30 – 28:15

Demo of declarative management

28:15 – 40:40

Features related to entity based development, storing content of any type, and SQL Server support for location-aware data

40:40 - 46:30

Demo of application & SQL Server integration with Virtual Earth

46:30 – 54:25

Data-warehouse and BI related features

54:25 – 58:20

Demo that did not work, but restarted at 59:05

59:05 – 1:05:00

Demo on how Dundas charts (will be part of SQL 2008) could improve your SSRS reports

In case you did not attend TechEd 2007, many presentations from that conference are now available as webcasts on http://www.microsoft.com/events/default.mspx.

SYSK 347: How-To Keep Historical/Audit Data Accurate

irenak — Thu, 10 May 2007 16:01:00 GMT

I’m sure you’ve heard this saying many times before – “the hard disk space is cheap now – save everything”… With that, many companies create paired (or buddy) tables for all transactional data, i.e. for every transactional table (i.e. not lookups, a.k.a. domain values) you’d create another one with same columns plus (usually) four more – AuditRecordID, Action (insert, update, delete), AuditTimeStamp, and AuditSQLUser (see SQL script below).

So, to make sure that anytime you insert, update or delete a record, a corresponding record is inserted into the “buddy” table, I prefer to use triggers. For security reason, I also have a trigger on the audit (“buddy”) table that prevents users from deleting or modifying data in it.

Below are such triggers for the insert and update actions (delete would be virtually same as insert):

set ANSI_NULLS ON

set QUOTED_IDENTIFIER ON

CREATE trigger [Table1_InsertTrigger]

on [dbo].[Table1] for insert

begin

insert into dbo.Table1_Audit

select 'i', getdate(), suser_name(), inserted.*

from inserted;

end

CREATE trigger [Table1_UpdateTrigger]

on [dbo].[Table1] for update

begin

-- NOTE: the commented out output clause below is not usable due to

-- having triggers on the audit table:

-- Msg 331, Level 16, State 1, Procedure Table1_UpdateTrigger, Line 7

-- The target table 'dbo.Table1_Audit' of the OUTPUT INTO clause cannot have any enabled triggers.

-- Make sure the timestamp is correct

update dbo.Table1

set LastUpdatedTimeStamp = getdate()

where pk_id in (select pk_id from inserted)

-- output 'u', getdate(), suser_name(), inserted.* into dbo.Table1_Audit;

-- Workaround for the output clause

insert into Table1_Audit

select 'u', getdate(), suser_name(), inserted.*

from inserted

end

set ANSI_NULLS ON

set QUOTED_IDENTIFIER ON

CREATE trigger [Table1_Audit_Trigger]

on [dbo].[Table1_Audit] instead of update, delete

begin

raiserror (N'Audit tables are insert-only', 16, 1);

end

You could go even further and create a DDL trigger that prevents users from dropping your audit tables – the example is provided in SQL Server 2005 Books Online (just look for DDL triggers).

Script to create Table1:

USE [YourDatabase]

SET ANSI_NULLS ON

SET QUOTED_IDENTIFIER ON

CREATE TABLE [dbo].[Table1](

[pk_id] [bigint] IDENTITY(1,1) NOT NULL,

[Data1] [nvarchar](128) COLLATE SQL_Latin1_General_CP1_CI_AS NULL,

[LastUpdatedTimeStamp] [datetime] NOT NULL CONSTRAINT [DF_Table1_LastUpdatedTimeStamp] DEFAULT (getdate()),

CONSTRAINT [PK_Table1] PRIMARY KEY CLUSTERED

(

[pk_id] ASC

)WITH (IGNORE_DUP_KEY = OFF) ON [PRIMARY]

) ON [PRIMARY]

Script to create Table1_Audit:

USE [YourDatabase]

SET ANSI_NULLS ON

SET QUOTED_IDENTIFIER ON

SET ANSI_PADDING ON

CREATE TABLE [dbo].[Table1_Audit](

[AuditRecordID] [bigint] IDENTITY(1,1) NOT NULL,

[Action] [char](1) COLLATE SQL_Latin1_General_CP1_CI_AS NOT NULL,

[AuditTimeStamp] [datetime] NOT NULL CONSTRAINT [DF_Table1_Audit_AuditTimeStamp] DEFAULT (getdate()),

[AuditSQLUser] [nvarchar](128) COLLATE SQL_Latin1_General_CP1_CI_AS NOT NULL CONSTRAINT [DF_Table1_Audit_AuditSQLUser] DEFAULT (suser_name()),

[pk_id] [bigint] NOT NULL,

[Data1] [nvarchar](128) COLLATE SQL_Latin1_General_CP1_CI_AS NULL,

[LastUpdatedTimeStamp] [datetime] NOT NULL,

CONSTRAINT [PK_Table1_Audit] PRIMARY KEY CLUSTERED

(

[AuditRecordID] ASC

)WITH (IGNORE_DUP_KEY = OFF) ON [PRIMARY]

) ON [PRIMARY]

SET ANSI_PADDING OFF

SYSK 345: How To Assure Correct LastUpdatedDateTime Value in Your SQL Tables

irenak — Tue, 08 May 2007 15:04:00 GMT

Often, developers add LastUpdatedDataTime (or LastUpdatedTimestamp) and LastUpdatedByUser columns to tables in relational databases.

And, as one might guess, these fields are often created to be not nullable with default values set by using SQL functions – getdate() and suser_name() – correspondingly.

The problem is that the default values only work on inserts, so, if the caller of your stored procedure (or TSQL statement) did not specify those columns in the UPDATE statement, then you’ll have stale (i.e. incorrect) data.

Here is what I do in such situation:

Create an after trigger on the table for update
Override those fields with the right values

For example:

CREATE trigger [Table1_UpdateTrigger]

on [dbo].[Table1] for update

begin

-- Make sure the timestamp and SQL user name are correct

update dbo.Table1

set LastUpdatedTimeStamp = getdate(),

LastUpdatedBySQLUser = suser_name()

where pk_id in (select pk_id from inserted)

end

NOTE: If your SQL user is not same as the application end user, you may want to have two LastUpdatedBy columns – LastUpdatedBySQLUser and LastUpdatedByAppUser (must be passed in from the tier where the end user identity is known).

SYSK 338: What SQL Queries Are Currently Running?

irenak — Fri, 27 Apr 2007 16:34:00 GMT

That’s easy -- just run the SQL statement below…

select session_id, request_id, start_time, status, cpu_time, er.total_elapsed_time,

substring(st.text,

(er.statement_start_offset/2) + 1,

((case er.statement_end_offset when -1 then datalength(st.text) else er.statement_end_offset end

- er.statement_start_offset)/2) + 1) as query

from sys.dm_exec_requests er

cross apply sys.dm_exec_sql_text(er.sql_handle) st

SYSK 333: What Query Plans Are Cached in SQL Server?

irenak — Fri, 20 Apr 2007 15:04:00 GMT

It’s well known that, to improve performance, SQL Server caches query execution plans in memory. But would you like to know what query execution plans are currently in the cache?

The following query will yield that information:

select substring(st.text,

(qs.statement_start_offset/2) + 1,

((case qs.statement_end_offset when -1 then datalength(st.text) else qs.statement_end_offset end

- qs.statement_start_offset)/2) + 1) as query,

qs.execution_count, qs.last_worker_time, qs.max_worker_time, qs.last_execution_time

from sys.dm_exec_cached_plans cp

inner join sys.dm_exec_query_stats qs on cp.plan_handle = qs.plan_handle

cross apply sys.dm_exec_sql_text(qs.sql_handle) st

On a different, but related note, to clear up procedure cache from SQL Server, run DBCC FREEPROCCACHE command.

SYSK 328: The “Hidden” System Database in SQL 2005

irenak — Fri, 13 Apr 2007 15:07:00 GMT

If you just open SQL Server Management Studio and expand the System Databases node, you’ll see four databases – master, model, msdb and tempdb (just like in SQL 2000).

However, SQL Server 2005 has a new system database called the ‘Resource’ database, which contains all the system objects (see select * from sys.objects), which appear in the INFORMATION_SCHEMA of each database but are actually stored in the Resource database. No user data or user metadata is stored in the Resource database.

The Resource database is read only for users, i.e. you can’t insert or update data at will.

I’m not aware of the reason, but baking up master database doesn’t automatically backup the Resource database (to me that would make sense). Moreover, you cannot backup the Resource database by using BACKUP DATABASE statement. Instead, you should copy it as a file to your backup share.

The physical file names of the Resource database are mssqlsystemresource.mdf and mssqlsystemresource.ldf. By default, these files are located in <drive>:\Program Files\Microsoft SQL Server\MSSQL.1\MSSQL\Data\. Each instance of SQL Server has one and only one associated mssqlsystemresource.mdf file, and instances do not share this file. In a cluster, the Resource database exists in the \Data folder on a shared disk.

Source: http://msdn2.microsoft.com/en-us/library/ms190940.aspx

Special thanks to Saleem Hakani for bringing this information to my attention!

SYSK 312: Why You Should Not Change SQL Server Service Account Using Services Management Console

irenak — Thu, 22 Mar 2007 16:03:00 GMT

The SQL Server documentation states that it’s strongly recommended to use SQL Server Configuration Manager and not Control Panel\Administrative Tools\Services (services.msc) when changing SQL Server or SQL Server Agent service account. The question is – why?

The reason is that when SQL Server service account is configured with a Windows NT account, SQL Server sets various Windows user rights and permissions on several files, folders, and registry keys. However, when you use Services Console to set SQL Server Service account, these permissions are not set and you may run into serious issues due to lack of proper security settings.

If you’ve already made changes to ‘Log on as’ setting on MSSQLServer or SQLServerAgent services using the Services Management Console, follow the steps below to manually change the security rights:

Give full control on the following registry keys and sub-keys to the Windows user account selected as the startup security context for SQL Services:

· HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SQL Server\90

· HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SQL Server\<MSSQL.x>

Give full control on the following file folder:

· \Program Files\Microsoft SQL Server\<MSSQL.1>\MSSQL

Special thanks to Saleem Hakani for this information!

SYSK 300: Why TABLESAMPLE Is Not Returning Specified Number of Rows

irenak — Mon, 05 Mar 2007 16:04:00 GMT

One of the new features in SQL 2005 is the clause that allows you to get a random set of rows. The syntax supports specifying either a percentage or a number of rows to return:

TABLESAMPLE [SYSTEM] (sample_number [ PERCENT | ROWS ] ) [ REPEATABLE (repeat_seed) ]

For example, using AdventureWorks sample database, you can issue queries like this:

SELECT * FROM Production.Product TABLESAMPLE SYSTEM (10 PERCENT)

SELECT * FROM Production.Product TABLESAMPLE SYSTEM (100 ROWS)

Interestingly, while the Production.Product table has 504 rows, the results of the two queries above when executed multiple times varies from returning as few as 55 rows to 175 rows, in my tests.

The reason for this is explained in SQL 2005 documentation:

“It [TABLESAMPLE command] generates a random value for each physical 8 KB page in the table. Based on the random value for a page and the percentage specified in the query, a page is included in the sample or excluded. Each page that is included returns all rows in the sample result set. For example, when specifying TABLESAMPLE SYSTEM 10 PERCENT, SQL Server returns all the rows on approximately 10 percent of the specified table's data pages. If the rows are evenly distributed on the pages of the table, and if there is a sufficient number of pages in the table, the number of rows returned should approximate the sample size requested. However, as the random value generated for each page is independent of the values generated for any other page, it is possible that a larger, or smaller, percentage of pages than requested are returned. The TOP(n) operator can be used to limit the number of rows to a given maximum.”

NOTE: by specifying a random number generator seed, you can assure that you get same resultset every time. To do that, use REPEATABLE(yourseed) clause:

SELECT * FROM Production.Product TABLESAMPLE SYSTEM (100 ROWS) REPEATABLE (83746)

Beware: this is not like the repeatable read isolation level. If another user makes changes to the data in the table, you will not get back the exact same rows. It is only true for a given "version" of the table.

Finally, another way of getting random data is by ordering the table using NEWID() as follows:

SELECT TOP(100) * FROM Production.Product ORDER BY NewID()

Special thanks to Saleem Hakani for this post!

SYSK 295: Difference between password handling in SQL 2000 and SQL 2005

irenak — Mon, 26 Feb 2007 16:10:00 GMT

Did you know that in SQL Server 2000, the engine used to maintain two copies/versions of each SQL Server login password. One contained the actual password supplied by the user, and the second one was the password converted by SQL Server to all uppercase letters.

This was done to speed up case-insensitive validation of passwords, as SQL 2000 allowed users to login using passwords in uppercase, lowercase (or combination). Although this behavior is convenient for many users, it makes password-guessing attacks easier by reducing the number of possible passwords.

With SQL Server 2005, SQL Server stores only one and original copy of the password. A password entered by a user must match the password stored on the server. If the password does not match the password stored on SQL Server, the login fails and user is denied access to SQL Server.

If the precise case of the password characters is forgotten, the password must be reset by using

ALTER LOGIN LoginName WITH PASSWORD = 'NewPassword';

command.

Special thanks to Saleem Hakani for this post!

SYSK 291: SQL 2005 SP 2 is Available and Comes with Good News About Licensing Costs

irenak — Tue, 20 Feb 2007 16:18:00 GMT

With SQL Server 2005 SP2 released to the web on Feb. 19, 2007, SQL Server Enterprise Edition customers, and I quote, “only need to purchase one license per physical processor regardless of the number of virtually deployed instances. “

In other words, “when a customer licenses all of the processors in a server, they can run an unlimited numbers of SQL Server instances on an unlimited number of virtual operating system environments, at no additional licensing cost. “

SQL Server 2005 Service Pack 2 download location: http://www.microsoft.com/sql/sp2.mspx

What’s new in SQL 2005 SP2: http://download.microsoft.com/download/2/b/5/2b5e5d37-9b17-423d-bc8f-b11ecd4195b4/WhatsNewSQL2005SP2.htm

SQL Server 2005 Service Pack 2 press release: http://www.microsoft.com/presspass/press/2007/feb07/02-19SQLSP2PR.mspx

SYSK 287: Custom C#/VB (CLR) Types Used in SQL Server

irenak — Wed, 14 Feb 2007 17:21:00 GMT

Did you know that you can use custom CLR types in Microsoft SQL Server by adding [Serializable, Microsoft.SqlServer.Server.SqlUserDefinedType] attribute? And, yes, you can implement and use methods – see Microsoft.SqlServer.Server.SqlMethod attribute…

Here are some good links for more information:

How to create a CLR-based user defined type:

http://msdn2.microsoft.com/en-US/library/ms131106.aspx

http://msdn2.microsoft.com/en-US/library/ms131079.aspx

Using CLR-based types:

http://msdn2.microsoft.com/en-US/library/ms131076.aspx

http://msdn2.microsoft.com/en-US/library/ms131086.aspx

http://msdn2.microsoft.com/en-US/library/ms131080.aspx

http://msdn2.microsoft.com/en-US/library/ms131055.aspx