Question

Here’s the scenario: there is a Software table (PK = SoftwareID) and an associated Release table (PK = [SoftwareID,Version]).

A release can be major or minor, release type is identified by Release.ReleaseType (‘MAJ’, ‘MIN’).

A release is also characterized by a date: Release.ReleaseDate.

Software is partitioned into categories, identified by Software.CategoryID.

Problem: need an effective T-SQL query to list all software pieces of a certain category and having the first major release date falling inside a given interval, delimited by @DateFrom, @DateTo. The only columns needed in the final resultset are SoftwareID and ReleaseDate.

This is not the real case scenario but I formulated it this way to be easier to understand. In the real case the table Release would have around 10 million records and the table Software around 1 million. I came up already with a solution but it’s quite slow and I feel the experts around here might find something better.

Here’s my slow solution:

select  s.SoftwareID, min(r.ReleaseDate)
from
    Software s inner join Release r on (s.SoftwareID = r.SoftwareID)
where s.CategoryID = @Category
      and r.ReleaseType = 'MAJ'
group by
    s.SoftwareID
having
    min(r.ReleaseDate) >= @DateFrom
    and min(r.ReleaseDate) < @DateTo

Thanks.

Answer 1

Your query is good.

You may want to make sure you have proper indexes for your query:

• in “MSSQL Management Studio”: “Query”->”Analyze Query in Database Engine Tuning Advisor”, then follow the “white rabbit” (preferred)
• in “MSSQL Management Studio”: “Query”->”Include Actual Execution Plan”. Then run the query, and look if the execution plan suggests new indexes.