I’ve moved this blog over to Blogger.  The new address is now: http://www.mcsdeveloper.com

I recently ran across a problem while trying to register a custom ASP.NET control on a themed page.  First I modified web.Config to register an assembly for use on any page or control in the site (take note of the namespace):

<system.web>
   <controls>
      <add namespace=”Insight.Web.CustomControls” tagPrefix=”custom” assembly=”Insight.Web”/>
   </controls>
</system.web>

Next, I created a theme named Insight and registered it site-wide in web.Config:

<system.web>
   <pages masterPageFile=”~/default.Master” styleSheetTheme=”Insight“>
</system.web>

Note that first segment of the namespace (Insight) matches the name of the Theme.  This will cause an error:

“Type Insight.Web.CustomControls.GridView is not defined”.

The type was defined just fine thank you very much, ASP.NET gets confused because part of the type name matches the name of the theme.  But you can fix this easily by just changing the name of the theme from Insight to something else (like InsightTheme):

<system.web>
   <pages masterPageFile=”~/default.Master” styleSheetTheme=”InsightTheme“>
</system.web>

And presto! No more error.  It’s best to set the default theme once in web.Config instead of on each page, making problems like this much easier to fix!

At BI Training, we glossed over using PPS to create trend charts. I decided to look into this in more detail and blog about it. Below is a basic demo of the steps required to use your existing Analysis Services KPIs within PPS to create trend charts.

Setup environment:

Download the SQL Server Samples from CodePlex.

Open the AdventureWorks sample solution located at: C:\Program Files\Microsoft SQL Server\90\Tools\Samples\AdventureWorks Analysis Services Project\Standard\Adventure Works.sln

You’ll notice there are a few KPIs in the Adventure Works cube (Growth in Customer Base, etc…). Deploy and Process the project (by right-clicking and choosing Deploy). FYI, the KPIs should look something like this in BIDS. We will be looking at how to import one or more of these KPIs [from Analysis Services] into Performance Point Server 2007 to use for trend analysis.

Create the datasource

• Open Dashboard Designer and add a new datasource. Set it up as follows:
  • From the “Multidimensional” category, choose the “Analysis Services” template.
  • Name the data source “AdventureWorksDS”
  • Check the checkbox labeled “Grant read permissions to all authenticated users” and click Finish
  • Setup remaining properties per the below screen shot:
    

Create a scorecard

• Add a new Scorecard, and choose the “Analysis Services” template from the “Microsoft” category. Be sure the “Use wizards to create scorecards” box is checked. Click OK.
• Type “TrendAnalysisSC” as the Name, and check the checkbox labeled “Grant read permissions to all authenticated users”. Click Next.
• Select the “AdventureWorksDS” data source and click Next.
• Select the 2^nd radio button labeled “Import SQL Server Analysis Services KPIs.
• Select the “Internet Revenue” KPI and click Next.
• Click Next again to skip the “Add measure filters” step.
• Check “Add column members” and select the “Date.Date.Fiscal” dimension.
• Now we need enough historical data to produce a good forecast. In “select members for columns”, select each quarter for 2002 and 2003. It should look like this:

• Click Finish. Then click Close.
• Now click “Publish All” in the ribbon area. This is very important. If you don’t do this, the scorecard may not show up during the following steps.
• Now we need to create a trend chart based on the TrendAnalysisSC Scorecard.
• To do this, add a new report. When prompted for report type, choose “Trend Analysis Chart”. Click OK.
• Name the report “TrendAnalysisReport”, and check the checkbox labeled “Grant read permissions to all authenticated users”.
• In the “Select a scorecard” step, choose “TrendAnalysisSC”.
• In the “Select the KPIs” step, choose “Internet Revenue”. Click Finish. Then click Close.
• You should see something like this:

• Note: The above is a chart of the actual KPI values, not the forecasted values. To see forecasted values, enter 2 in the “Forecast Period” box and hit Tab to leave the field. The chart will refresh and highlight the forecasted section red. Because we entered a value of 2, and we are at the “Quarter” level, the values for the next two quarters are forecasted. The below graph is how it should looknow:

• Note that three series (lines) are displayed. Click the “Show/hide Legend” button in the toolbar to see what these are. There will be a series for each element of the KPI (Value, Goal/Status, Trend).
  Note: if you don’t see the toolbar, right-click the chart and select “Toolbar”.
• So what if you are asked to hide everything except the “Value” series? Read on…

How to remove one or more series from the trend chart

• In the “Workspace Browser”, click on the “Internet Revenue” KPI you created earlier. You should see a grid with three rows. Delete the “Goal and Status” row and the “Trend” row. The only remaining row is now “Value”.
• Click “Publish All” and then click TrendAnalysisReport. Nothing has changed. Now click the refresh button. Still, nothing changes. Also feel free to go into the scorecard and delete all of the “Goal and Status” and “Trend” columns. The report still shows all three series. Guess what… the report is now a completely separate entity from the KPI and the Scorecard; so modifying these will do nothing to change the report. To remove a series from the report, you have a few options:
  • (Bad solution) Modify the scorecard, delete the report, then recreate the report based on the modified scorecard.
  • (Good solution) Right-click the chart area of the report and choose “Field List…”. You’ll see something like the below. Right-click “Goal and Status” and choose “Delete” to remove that series from the chart.
    

You should now see only the Value series:

FYI: Performance Point Creates Temporary Mining Models

• The forecasted data points are created by Analysis Services, but in this case, a temporary session-scoped Mining Model is created for you (on-the-fly) by PPS. To prove this to yourself, fire up SQL Profiler, then start a trace using Analysis Services and the Adventure Works cube. Now run the report (you may have to change the “Forecast Period” again to see results. From the confusing results of the trace, below are a few highlights / observations:

1: The below MDX Query is generated. I suspect the results of this are used as @InputRowset (keep reading) to train the Mining Model:

WITH

MEMBER [Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_12af2876-ed38-4404-a753-c5abe35331cc] as ‘(KPIGoal(“Internet Revenue”))’

MEMBER [Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_12af2876-ed38-4404-a753-c5abe35331cc] as ‘([Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_12af2876-ed38-4404-a753-c5abe35331cc])’

MEMBER [Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_bb983dd8-365c-45ff-846a-b0d4f47bfee9] as ‘(KPIStatus(“Internet Revenue”))’

MEMBER [Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_bb983dd8-365c-45ff-846a-b0d4f47bfee9] as ‘([Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_bb983dd8-365c-45ff-846a-b0d4f47bfee9])’

MEMBER [Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_3829f331-b17f-4647-806c-8370d2266c56] as ‘(KPITrend(“Internet Revenue”))’

MEMBER [Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_3829f331-b17f-4647-806c-8370d2266c56] as ‘([Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_3829f331-b17f-4647-806c-8370d2266c56])’

MEMBER [Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_26aa5e8c-4b17-44f2-b6f2-fb6adcf6b662] as ‘(KPITrend(“Internet Revenue”))’

MEMBER [Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_26aa5e8c-4b17-44f2-b6f2-fb6adcf6b662] as ‘([Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_26aa5e8c-4b17-44f2-b6f2-fb6adcf6b662])’

MEMBER [Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_fc0ccba1-69e1-4ce0-85ba-0dc0fe814c4d] as ‘(KPIValue(“Internet Revenue”))’

MEMBER [Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_fc0ccba1-69e1-4ce0-85ba-0dc0fe814c4d] as ‘([Measures].[CUSTOM_a7ba172a-ce41-4d38-b573-c7bed75a9477_fc0ccba1-69e1-4ce0-85ba-0dc0fe814c4d])’

SELECT { [Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_12af2876-ed38-4404-a753-c5abe35331cc],

[Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_bb983dd8-365c-45ff-846a-b0d4f47bfee9],

[Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_3829f331-b17f-4647-806c-8370d2266c56],

[Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_26aa5e8c-4b17-44f2-b6f2-fb6adcf6b662],

[Measures].[a7ba172a-ce41-4d38-b573-c7bed75a9477_fc0ccba1-69e1-4ce0-85ba-0dc0fe814c4d]} ON AXIS(0),

{([Date].[Fiscal].[Fiscal Quarter].&[2003]&[1]),

([Date].[Fiscal].[Fiscal Quarter].&[2003]&[2]),

([Date].[Fiscal].[Fiscal Quarter].&[2003]&[3]),

([Date].[Fiscal].[Fiscal Quarter].&[2003]&[4]),

([Date].[Fiscal].[Fiscal Quarter].&[2004]&[1]),

([Date].[Fiscal].[Fiscal Quarter].&[2004]&[2]),

([Date].[Fiscal].[Fiscal Quarter].&[2004]&[3]),

([Date].[Fiscal].[Fiscal Quarter].&[2004]&[4]),

([Date].[Fiscal].[Fiscal Quarter].&[2005]&[1]),

([Date].[Fiscal].[Fiscal Quarter].&[2005]&[2])} ON AXIS(1)

FROM [Adventure Works]

2: PPS creates a mining model in Analysis Services (note the SESSION keyword, which causes the model to be removed when the connection closes):

CREATE SESSION MINING MODEL MiningModel_0( [KeyTime] LONG KEY TIME ,[Key] LONG KEY ,[Column_2] DOUBLE CONTINUOUS PREDICT_ONLY) USING Microsoft_Time_Series

3: Training data is inserted into the model. This causes the model to be processed / trained.

INSERT INTO MiningModel_0 ( [KeyTime] ,[Key] ,[Column_2]) @InputRowset

4: Results from DMX query are returned. Note that two forecasted periods are requested:

SELECT FLATTENED (SELECT $TIME, [Column_2] from PredictTimeSeries([Column_2], 2) ) FROM [MiningModel_0]

That’s it. Hopefully this gives you a good general idea of how to create trend charts in Performance Point Server 2007 using your existing KPIs from Analysis Services, and of what’s going on behind the scenes.

Ever get the below error when trying to connect to Analysis Services?  From the looks of things at forums.microsoft.com, lots of folks do. 

The problem is the account that the “SQL Server Browser Service” uses to authenticate, which by default is the account selected for services during setup.  But this account needs to have administrator rights.  So if the “Network Service” account is used, the quick fix is to change the SQL Server Browser Service to run as “Local System” instead.  Have a look at this Books Online article for details on the minimum rights required for the browser service to work properly – you’ll want this account to have the least permissions required in a production environment.

But according to Microsoft: “To work properly, the SQL Server Browser service must run under a security account that has local administrator rights, such as the local system account.” (Pasted from http://www.microsoft.com/technet/prodtechnol/sql/bestpractice/CISQL2005ASCS.mspx)

This seems a bit ambiguous to me.  The books online article recommends using minimum permissions, but the technet article says you have to use an admin account.  On my laptop, the browser service appears to have all of the recommended rights, but will not let me connect until I run as Local System.  Strange but true.

Note: The SQL Server Browser Service in (SQL Server 2005) is only necessary when using named instances or a side-by-side installation with SQL Server 2000.  Otherwise, SQL Server setup does not configure the browser service to start automatically.

I’m attending an Advanced BI Training course this week.  Below are some books and online resources recommended by the instructor.  

Books:

• The Rational Guide To Monitoring and Analyzing with Microsoft Office PerformancePoint Server 2007
• Business Intelligence with Microsoft® Office PerformancePoint Server 2007
• Information Dashboard Design
• Fast Track to MDX 
• The Microsoft Data Warehouse Toolkit: With SQL Server 2005 and the Microsoft Business Intelligence Toolset
• MDX Solutions
• Data Mining with SQL Server 2005

Web Sites:

• http://www.TDWI.org (The Data Warehouse Institute)
• http://blogs.msdn.com/performancepoint/
• http://www.perceptualedge.com/ 
• http://www.sqlserverdatamining.com/

Tools:

• BIDS Helper : “A Visual Studio.Net add-ins with features that extend and enhance the functionality of the SQL Server 2005 BI Development Studio. (BIDS)”

Attempting to install VS 2008 RTM on my PC earned me the following error message:

“the following component failed to install: Microsoft Visual Studio Web Authoring Component”

Prior to the 2008 RTM, I had installed both the Orcas beta and a separate green bits version from Microsoft.  I got the green bits version for a Microsoft project I worked on that required the latest unreleased version of Orcas.  So I was expecting a problem or two just cleaning up the old stuff. 

I also run Altiris SVS which lets you install applications into isolated virtual layers.  But apparently all of the layers (applications) in SVS need to be deactivated for the Web Authoring Component to install.  It’s a weird and obscure fix that I would never have figured out without this kb article: http://support.microsoft.com/kb/933568

Just go into the SVS admin tool and deactivate all layers. VS 2008 should install without a hitch. 

I had just activated a new feature on a WSS 3.0 site, and attempted to add one of its Web Parts to a page, when up came a helpful error message:

“The Web Part you attempted to add no longer exists in the Closed Web Parts Gallery.  Refresh your browser to update the Web Part Page and the Gallery.”

But the problem is that I attempted to add an Ajax web part to a page without first installing ASP.NET Ajax.  Daniel Larson said it best: “This is the WORST error handling ever.” This error message offers almost no help.  Daniel Larson’s blog post on the other hand, was a big help (many thanks)!

This brings to mind another infamous SharePoint error message that appears after modifying any aspect of a “ghosted” page with SharePoint Designer, which moves the page to an “unghosted” state (i.e., the page is moved from disk into the content database) and causes the page to be executed in Safe Mode.  If any inline scripts are in the page, you might see this friendly reminder:

“An error occurred during the processing of /SomeDir/SomePage.aspx. Code blocks are not allowed in this file.”

This message makes sense after you’ve seen it a few times, but isn’t very helpful if you’re new to WSS, and making even a trivial change breaks your page.  Modifying web.config to always allow server scripts to run is a quick and dirty fix for this one:

<PageParserPath VirtualPath=”/*” CompilationMode=”Always” AllowServerSideScript=”true” IncludeSubFolders=”true” />

But a better (more secure) approach would be to create a custom server-side control, and explicitly register it as safe in web.config :

<SafeControl Assembly=”MyApp, …” Namespace=”MyApp.Controls” TypeName=”*” AllowRemoteDesigner=”True” />

I suspect Microsoft is letting developers write their own error messages, which might explain why they read like code.  As long as people keep blogging their solutions though, I’m happy.

I’m working on a currency localization feature in a banking application, and working through different ways to parse currency strings.  One way to do this is to call Decimal.Parse(), passing in the currency string and an instance of NumberFormatInfo, which specifies how the string is formatted.  Here’s an example:

Parsing Swiss Franks from a string to a decimal:

NumberFormatInfo infoSwissFranc = new NumberFormatInfo();
infoSwissFranc.CurrencyDecimalSeparator = “,”;
infoSwissFranc.CurrencySymbol = “SFr.”;
infoSwissFranc.CurrencyGroupSeparator = ” “;
Console.WriteLine(Decimal.Parse(“SFr. 1 111 111 110,11″, NumberStyles.Currency, infoSwissFranc).ToString());

Here’s another example that parses the Swedish Krona format:

NumberFormatInfo infoSEK = new NumberFormatInfo();
infoSEK.CurrencySymbol = “kr”;
infoSEK.CurrencyGroupSeparator = “-”;
infoSEK.CurrencyDecimalSeparator = “,”;
Console.WriteLine(Decimal.Parse(“111-111-110-11 kr”, NumberStyles.Currency, infoSEK).ToString());

Oops, the 2nd snippet doesn’t work.  In this case, assigning a value of CurrencyDecimalSeparator is not good enough.  You must also assign the same value to NumberDecimalSeparator as well, or an exception will be raised.  Apparently, this is only necessary if CurrencyGroupSeparator is a decimal (CurrencyGroupSeparator =”.”)  and the currency symbol is at the end of the string instead of the beginning.  Although setting NumberDecimalSeparator may be required when parsing when parsing numbers using NumberStyles.Number, it should not be required when NumberStyles.Currency is specified.  I could be missing something, but it looks like a glitch in the .NET framework. 

After adding NumberDecimalSeparator (in red), this code will successfully parse the Swedish Krona format:

NumberFormatInfo infoSEK = new NumberFormatInfo();
infoSEK.CurrencySymbol = “kr”;
infoSEK.CurrencyGroupSeparator = “-”;
infoSEK.CurrencyDecimalSeparator = “,”;
infoSEK.NumberDecimalSeparator = “,”;
Console.WriteLine(Decimal.Parse(“111-111-110-11 kr”, NumberStyles.Currency, infoSEK).ToString());

I created this wallpaper for my desktop from this image, courtesy of our friends at Microsoft.    It’s sized for 1920 x 1200 resolutions, with a solid black border around the top/left to make room for desktop icons.  Enjoy! 

I like using free online mortgage calculators to get a quick idea of how much a new home will cost, or generally how much I can shorten the life of my existing mortgage by making overpayments.   I recently put this same functionality into an Excel spreadsheet, which offers several advantages over the online calculators:

1. Less Work: You never have to re-enter any information.
2. Privacy / Security: It’s all saved on your PC.
3. No internet connection required: Excel runs on your PC.
4. Convenience: No more google searches for: “mortgage calculator”.
5. Flexibility: Use this spreadsheet as a starting point, then customize or extend it yourself.
6. Accurate Overpayment Calculation: Enter specific overpayment amounts and dates.  Some online calculators allow you to enter one or more “one time” payments.   Others allow you to enter an overpayment amount, but they assume that this amount will be applied starting with the first payment.  Many homeowners will decide to start making overpayments after living in their home for several years.  This spreadsheet allows you to enter any number of “one time” payments, but there is also a cell for “Estimated Future Overpayments”.  This amount will be applied starting with the first payment after the current date, until the loan is paid off.  With the spreadsheet, you will see a more accurate amortization based on a combination of past and expected future overpayments.

 To download the mortgage calculator, click here, then keep reading for some basic instructions:

Step 1: Enter loan information.  Enter the Principal amount, Number of Payments (360 for a 30 year loan), etc.  Do not modify any field with a gray background (these are automatically calculated for you).  Note the “Estimated Future Overpayment” cell I mentioned earlier.  This amount will be added to any future payment amount.

                                                                               figure 1

Step 2: Go to the “Overpayments” worksheet and enter previous overpayment dates and amounts.  The spreadsheet will automatically figure out which payment date to apply these overpayments to.  Because most mortgage loans are simple interest, it’s very important that overpayments are accounted for in the amortization on the specific date of the overpayment.  The principal is reduced further by the overpayment amount, which reduces the amount of interest you owe in the next payment. 

                                           figure 2

Step 3:  That’s it.  Enjoy!   Your amortization should look something like the below image.  Note the gray background, which means these values were calculated for you.  Note that the last payment date is also calculated for you (see figure 1), and takes into account any overpayments (actual and estimated).

                                                                                                                                                                            figure 3