Introduction

This is a description of how to put together a report that tracks change to some data over time. This example will be a basic Product Backlog report.

For this work we will use Report Builder 3.0 and use the query editor in MS SQL Management Studio to write the SQL.

You need to have a way of writing, running, tuning, and testing the SQL for your reports. One way to do this is with SQL Management Studio; it provides good visibility of the structure of TFS databases and provides good tools for creating the SQL that will ultimately be used in the report. 

You could use the graphical UI or SQL text editor found in Report Builder;  but to tweak and develop the SQL you will most likely discover that you need to get more control and feedback than the Report Builder editor provides. The editor in SQL Server Management Studio has Intellisense which is useful. Another good alternative would be to use Visual Studio.

For this example report we will need to read data from a couple of  tables in the TFS Warehouse; specifically DimWorkitem and FactWorkitemHistory.

So we need to set up SQL Management Studio to connect to TFS SQL server; which when connected looks like the image on the right.

We can then select DimWorkItem and click ‘New Query’.

We’ll write some SQL that will generate an array of dates for the X axis and for each of those dates sum up the effort outstanding on that date. The full SQL is appended at the end of this article.

Writing the SQL for the report.

We need a start date and end date for the array of dates. We’ll simply get the minimum and maximum dates from the workitem data; so selecting from DimWorkitem we’ll set @vDataSetStartDate and @vDataSetEndDate as below. Also we’ll limit this example to the ‘Product Backlog Item’ workitem type. No effort is made here to restrict the data to a particular Project Collection or Team Project; so be aware that the volume of data returned could be quite large.

With the start and end date handy we can then populate an array of dates. We’ll put the dates into a temporary table called @tblDates. You could use select * from @tblDates to see the dates in this table (see the inset to the right); remember to remove the select * before using the SQL in a report.

For each of those dates in @tblDates we’ll get all the latest revisions of workitems up to and including the DayDate. For each date we store the data that will be used for filtering in the last step where we will sum up remaining effort based on workitems that are in the ‘not done’ state. We will need the WorkItemSK (which is the unique id for a revision) to join into the FactWorkitemHistory. The ID and the Rev columns are for reference and debug. Use select * from @tblDatesWithWorkitems to see the dates and data in this table; remember to remove it before using the SQL in the report.

The data we want requires joining DimWorkitem with FactWorkitemHistory; INNER JOIN on the WorkitemSK. There are rows in FactWorkitemHistory that are irrelevant; so filter them out using fwih.RecordCount > -1.

Finally filter by system state for the ‘not done’ workitems from the temporary table @tblDatesWithWorkitems and SUM up Effort for each day by grouping on DayDate. So for example on 2012-10-02 join each row that matches our state criteria to FactWorkitemHistory and sum up.

So that’s the data ready to go.

Importantly it is in the exact shape and format that we need for our report. An array of Dates with associated backlog of effort.

We could enhance this data with additional information about sprints, teams and velocity.

We could add filtering for Project Collection, Team Project, Iteration, Areas and the like. 

That will be a subject for another article!

Getting the data into a report. Creating a DataSet.

Open up MS SQL Server Report Builder. Close the wizard and go straight into the application. Create a Data Source pointing at the TFS SQL Server.

Add a data source that points at the TFS warehouse. Here I use the ‘Use a connection embedded in my report’ option with connection type ‘Microsoft SQL Server’ and name the connection TFS2012.

Create a dataset using the previously created data source.

In the Dataset Properties dialog open the Query Designer, select ‘Edit as Text’ and paste in the SQL that was previously created in SQL Server Management Studio. Run the SQL to check that it works ok. Although of course if you are using SQL Server Management Studio to develop this type of report SQL you will have already seen that it works; so you would not normally need to do this.

Add a table to the report so that you can view the data that is going to be charted. This is a useful way to verify that the data looks shipshape! The table can be deleted later. Lift click and drag DayDate and Effort into the table.

Add a chart.

Select the first Line type chart.

Then left Click drag DayDate onto the chart as a category group.

Then left Click drag Effort as a value.

Use the expression editor and edit ‘Effort’ to remove the Sum() and then OK to save the modified expression.

Run the chart.

That’s it. You have day by day data in a SQL Report.

• Summary.

• The approach taken here is a pattern used in many SQL Reports that require data to be charted against days.
• Use SQL to generate an array of dates into a temporary table. These dates should be limited by the date range of the data being charted. For each Date get the maximum revision for the workitems with change date falling before or on the Date. Use additional filters where appropriate to define the workitems; but remember the aim is to get the latest revision of each workitem for the Date. So for that date we can look at things as they were on the day; as we were looking at them through a Query run on that day. The latest revision may not have taken place that day; it most likely will be an earlier day. Finally SUM up by grouping the data by Date.

The complete SQL.