Question

For the following table :

Id    Timestamp    AssignedTo
1     2012-01-01   User1
2     2012-01-02   User2
3     2012-01-10   User3
4     2012-01-15   User1

what would be the general approach in MS SQL Server to calculating how many days/hours/minutes the entity was assigned to a specific user?

As a developer (with basic SQL knowledge), I’ve always opted for Functions or Stored procedures when doing these kind of queries (these kind = queries including results based on the difference between two rows in a set).

In this example, I’d have a cursor iterating over the items, keeping the previous one in a variable, and then calculating the difference. However, I’ve been told that this is a serious performance concern, especially on large sets.

EDIT: Sample results

User    TotalTime
User1   23d
User2   8d
User3   5d

i.e. the total time that an item was assigned to a specific user. User1’s TotalTime is 23 days because it has been assigned to him since 2012-01-15 (and it’s 2012-02-06 now), along with the first day it was assigned to him.

Answer 1

In this example I’d use a subquery within the select statement to get the end date for each row, then use this end date to get the time assigned to a specific user. The below is SQL Server syntax but the same principal can be applied whatever the RDBMS.

SELECT  AssignedTo, 
        SUM(DATEDIFF(DAY, TimeStamp, ISNULL(EndDate, GETDATE()))) [Days]
FROM    (   SELECT  ID,
                    [TimeStamp],
                    AssignedTo,
                    (   SELECT  MIN(TimeStamp)
                        FROM    [YOURTABLE] b
                        WHERE   b.TimeStamp > a.TimeStamp
                    ) [EndDate]
            FROM    [YOURTABLE] a
        )  a
GROUP BY AssignedTo