How to group data at granularity level hours and every 10 minutes?

Question

I have sales table imported from SQL server. The date columns are are captured at granularity level such as 30-12-2013 16:50:16. 
what is best way create master date table, and create relationship between the sales table and the master date table?
I am new in Power Pivot, and any complete detailed steps and links will be appreciated
Note: just clarify the requirements. The business wants to see how much sales a consultant sells in  every 10, 30 minutes during normal business hour.
Hope this help
jl

Answer 1

Split your field:
1) A date portion related to your date dimension at the day grain
2) A time portion related to your time dimension at a 10 minute grain.
In TSQL:
SELECT
,[Date] = CAST( <datetimefield> AS DATE) -- date with no time
,[Time] = CAST( '18991230 '
+ RIGHT('0' + DATENAME(Hour, <datetimefield>), 2) + ':'
+ RIGHT( '0' + CAST( (DATEPART(MINUTE, <datetimefield>) / 10) * 10 AS NVARCHAR(2)), 2) + ':'
+ '00' AS DATETIME)
Write this in your query to populate the fact table. This will give you a field holding just the date, and one holding the time at a 10 minute granularity.
A date table is trivial to produce in SQL or in Excel.
Here's a link for doing a very basic one in SQL Server.
A time table is trivial as well:
WITH TimeCTE ([Time]) AS
(SELECT CAST('18991230 00:00:00' AS DATETIME)
UNION ALL
SELECT DATEADD(MINUTE, 10, [Time])
FROM TimeCTE
WHERE [Time] < CAST('18991230 23:50:00' AS DATETIME)
SELECT * FROM TimeCTE OPTION(MAXRECURSION 0)
This gives you the beginnings of a dimension with time intervals every 10 minutes.
You can extend this table with TSQL functions or DAX calculated columns, whichever you find more convenient.
Then, you can join your <fact table>[Date] to <date dimension>[Date], and your <fact table>[Time] to <time dimension>[Time], and do all of your filtering on those tables.
Note: I have used a full datetime field for the time dimension above. This is because Power Pivot/Tabular only know datetime as a data type. If you want to add time to a date, the time portion must be recorded on 1899-12-30 to achieve the desired result.
When importing a TIME data type into the Tabular model, the field is assigned the date of processing, which is absurdly annoying.