If you have ever created Indexed Views before you most definately would have come across an error or two throughout the process.  Before you put this into the too hard basket consider the following scenarios where the benefits of fast query performance will be realised…

Applications that benefit from the implementation of indexed views include:

• Joins and aggregations of large tables
• Repeated patterns of queries
• Repeated aggregations on the same or overlapping sets of columns
• Repeated joins of the same tables on the same keys
• Combinations of the above

Indexed views are therefore likely to benefit in the following scenarios :

• Decision support workloads
• Data marts
• Data warehouses
• Online analytical processing (OLAP) stores and sources
• Data mining workloads

Ok lets look at some basics.   The first thing we need to do is create our view.  We can then create one or more indexes on our view starting with a clustered index.

To create the view we will use a similar syntax to creating a regular view except we will use the WITH SCHEMA BINDING Option see Schema Binding …

e.g.

CREATE VIEW dbo.vw_SomeView WITH SCHEMABINDING

AS

SELECT colKey, col2, col3

FROM dbo.Table1

Note that any tables mentioned in the SELECT statement of your view must be in the two part owner.object name format.  There are a heap of caveits that could bring you down at this point which would prevent you from creating the view.   One of which is you should not use deterministic functions.  E.g. DateAdd returns the same value for a set of parameters each time, whereas getdate() does not i.e don’t use getdate() in your view.    CONVERT(datetime, somedate) is non deterministic as it can return different results when different values are provided for the style parameter….consider CONVERT(datetime, somedate,103) as it is deterministic – see Using Convert Within an Indexed View.   Note in the example Create View statement above there is explicit specification of columns in the select clause (i.e. no select * …), the select is on a base table not other views and if joins were used, OUTER JOINS would be avoided. The full design considerations for indexed views are outlined here http://msdn.microsoft.com/en-us/library/ms191432.aspx.

Once we have successfully navigated the design guidelines and have successfully created our view, now it is time to create the indexes.  We create the index as we would on a table.  The first index we need to create on a view is a unique clustered index.  We can then create additional none clustered indexes as required.

CREATE UNIQUE CLUSTERED INDEX ci_vw_Table1 ON dbo.Table1 (colKey)

Finally Aaron Bertrand runs through a practical example of using Indexed Views and discusses the good and dark sides of indexes views in this nice example Redundancy is No Evil.

Creating a data model that fits the business requirements like a glove might do for now, but the business are great at thinking up new ways of categorising their data.   Top items, these items, those items, those items excluding top items, the list goes on. Whilst some I.T. folk might view business requests as “annoying”, we in the B.I. space love seeing out business users take hold of our collaborations and running with them.

The following snippet demonstrates (using Adventureworks), how to group product lines into 3 custom groupings : Top 2 items, All Other Items, and All Items.  These techniques show you how to include specific line items of interest in your report, whilst grouping the remaining items you are not so interested in, into one total e.g. Other Products … From the Martin Mason Blog …

WITH SET [Top Items] AS

Generate(

{ [Product].[Category].[Category].Members },

TopCount(

EXISTING [Product].[Product].[Product].members,

2,

( [Measures].[Reseller Sales Amount] )

)

)

MEMBER [Product].[Product].[All].[All Other Products] AS

Aggregate( { EXISTING { [Product].[Product].[Product].Members } - [Top Items] } )

SELECT        {

[Measures].[Reseller Sales Amount]

} ON COLUMNS,

{

{ [Product].[Category].[Category].Members }

* {

[Top Items],

[Product].[Product].[All Other Products],

[Product].[Product].[All]

}

} ON ROWS

FROM        [Adventure Works]

Alternatively lets say the business wants to see all of the products that are listed in the current 20% catalogue sale (20% off all Accessories and Products), we can display the line items for all products in the Accessories and Products categories, then show all other items grouped into one total…

WITH SET [Catalog Accessory Products] AS

'Filter([Product].[Product].[Product].Members,

(InStr(1,[Product].[Category].CurrentMember.Name, "Accessories") <> 0)

)'

SET [Catalog Clothing Products] AS

'Filter([Product].[Product].[Product].Members,

(InStr(1,[Product].[Category].CurrentMember.Name, "Clothing") <> 0)

)'

MEMBER [Product].[Product].[All].[All Other Products] AS

Aggregate(

{ EXISTING

{ [Product].[Product].[Product].Members }

- {[Catalog Accessory Products] + [Catalog Clothing Products]} } )

SELECT        {

[Measures].[Reseller Sales Amount]

} ON COLUMNS,

NONEMPTY({

{ [Product].[Category].[Category].Members }

* {

[Catalog Accessory Products],

[Catalog Clothing Products],

[Product].[Product].[All Other Products]

}

}) ON ROWS

FROM        [Adventure Works];

Those turning their hand at MDX filter queries and doing custom member rollups filters …..(exhale) be aware the Performance Point 2010 MDX editor has a strict requirement on formatting which excludes the use of WITH statements before your SELECT as identified here MSDN Reference.

Ron Davis provides a good You Tube clip demonstrating the use of the MDX Filter feature Ron Davis MDX Filter Clip.

If you do try to perform some arithmetic on this column directly, you will most likely get the error … “Error converting data type varchar….”.

Now there is nothing like a clean datamart to store this data, however the following example demonstrates how to perform this function in an adhoc query.   The code sample searches the varchar column aptly named “Varchar Actual ($)” and replaces any symbols containing the characters $ , – and the space character with empty string…allowing the string to be successfully cast to a number…

CAST(REPLACE(REPLACE(REPLACE(RTRIM(LTRIM

([Varchar Actual ($)])),'$',''),',',''),'-','') as float)

AS [Float - Actual ($)]

Reference …

Social MSDN Thread

Cast and Convert

Cast and Convert Basics

Some REPLACE basics with Bru ! Have a Cuppa with Bru

To make matters worse the error is in relation to an attribute that is not part of the hierarchy you created.  You have checked and you don’t have any null values.

One of the conditions that can contribute to this error is when you have trailing special characters in the offending column such as char(13) or char(10).   The trimming property of the key column for the failing attribute is set to RIGHT by default.   So you will need to either trim the special characters out of your database, ETL or through the DSV.  Setting the trimming property to NONE may get you over the line in some circumstances but this is not foolproof and not a good long term strategy.

Bottom line – check your data for char(13) and char(10) characters, perhaps test the length of values in the offending column to make the problem easier to visualise.

SELECT col1, LEN(col1) AS Col1Length

FROM TableName WHERE col1 LIKE 'Value%' ORDER BY col1 

Additional reading …

Official BI Twibe Blog

ALTER TABLE dbo.TableName ALTER COLUMN ColumnName varchar(10) not null

But wait, there’s more !  To determine any other varchar(7) columns that could potentially be impacted you could issue a query against your database like the following, then apply similar code above to change those columns …

SELECT TABLE_CATALOG, TABLE_SCHEMA, TABLE_NAME, COLUMN_NAME, IS_NULLABLE, DATA_TYPE, CHARACTER_MAXIMUM_LENGTH
 FROM INFORMATION_SCHEMA.COLUMNS
 WHERE DATA_TYPE = 'varchar' and CHARACTER_MAXIMUM_LENGTH = 7

But wait, there’s still more !  If this particular column is involved in any SSIS packages you will need to refactor those packages.   If your SSIS package is stored in the MSDB database you may have some luck performing a SSIS Package Object Search, otherwise there are various 3rd party tools you can use to scan through your SSIS package and allow object searching.  I touched on this in the blog post PopBI DTS SSIS Object Search - there are many tools to facilitate this.

For any cube impacts you can install and then utilise the BIDS helper (Data Type Discrepancy Check).

For subtle data type changes such as the above example, your Reporting Services reports may still run without  error, however you may need to adjust the width of a column for example.

Ok now you can have the steak knives …

A link to the upgrade advisor can be reached here that includes the links to download the tool SQL Server 2008 R2 Upgrade Advisor Link - ordinarily this would be available in the SQL 2008 R2 feature pack.

Running the upgrade advisor is simple and allows you to select and analyse only the components you are thinking about upgrading.  For example if you only want to upgrade Reporting Services, then only select the Reporting Services option.  Keep in mind, for Reporting Services, you must install and run the tool locally on the Reporting Services server, whereas for the other components (SQL Server, Analysis Services, etc) you can analyse from a client workstation that has upgrade advisor installed.  Any issues can be viewed later using the same upgrade Advisor tool but using the Report Viewer option.

The link to the current SQL 2012 Release Candidate Upgrade Advisor is here SQL Server 2012 Upgrade Advisor Link.

I also recommend running the SQL Server Best Practices Analyser to help identify risk areas that might be better sorted out before you plan to do the upgrade.  The link to the Best Practices Analyser can be found here SQL Server 2008 R2 Best Practices Analyser Link.  BPA also performs some O.S. level checks so is recommended to install and run on the actual SQL Server you intend to analyse.  It is also works on SQL 2008 instances.

Currently, the SQL 2012 Best Practices Analyser defers to the SQL 2008 R2 version SQL Server 2012 Upgrade Quick Sheet.

Hope this helps.

Access to active directory data can be retrieved as simply as a linked server configuration and data access using OpenQuery. Brendan Tompkins provides a simple guide to achieving this but keep in mind the importance of getting the adsdatasource and port correct.

Brendan Tompkins A.D. Query Example

Some Ldap query examples are provided here – see Examples.

If linked servers are not your thing or you have problems accessing through the linked server then there is a tool that is installed with the Lightweight Directory Services Role for Windows 2008 called “csvde.exe” see CSVDE. This tool provides a command line interface to the active directory server to extract data in csv format using various object queries.

If you are using a non Microsoft Ldap v3 compliant directory server keep in mind the csvde.exe tool will still work, however you may run into problems. For a start rows may be paged in 500 row lots and csvde cannot page through the data automatically. Secondly, the columns may return to the csv file out of order when returning from the LDAP v3 compliant LDAP server (note this is unrelated to the –l behaviour as design which can return columns out of order if you specify specific attributes in your query against a Microsoft Active Directory Server). As a workaround you can get a Powershell guru like Troy Farrell to write a script to rearrange the data in the csv file.

There are third party tools available for a fee that allow you to interrogate and create command line extracts for LDAP v3 directory servers however be sure to test these tools thoroughly to handle the issues around data paging and out of order columns.

For non Microsoft LDAP servers that utilise single source of the truth for the directory in an  ODBC or OLDEB database, you also have the option of bypassing all of the above and extracting that data directly from the LDAP database using SQL Server Integration Services.

Happy A.D.

The expression to use to refer to the value inside a textbox within a report is …

ReportItems!textbox1.Value

Although, we are all good B.I. developers and would never leave our textbox named as textbox1 *wink* ….

Note if you are using the MDX Query Designer in SSRS you may find that editing your dataset queries effectively replaces any upstream datasets you just changed to the CHILDREN function back to ALLMEMBERS.  What the ??

As a workaround, you will need to jump into code view for your report, locate each dataset and under the datasets QUERY element you will need to add the <rd:SuppressAutoUpdate> element and set this to true.  This will effectively give the dataset immunity from being automatically updated.  Example …

<Query>
...
<rd:SuppressAutoUpdate>true</rd:SuppressAutoUpdate>
</Query>

Alternatively simply executing the parameter dataset is supposed to automatically set the <rd:AutoGenerated> element to FALSE (which has the same effect) however I had no luck here.

Discussed here Connect MDX Query Designer Bug.