Question

I am working on a survey application. I have a sql server database holding upto 3000 survey questions.
I call database for a set of 10 questions. I have to make sure that none of the of the questions are repeated for the longest time possible.
What is the best possible approach I can take here? Write a custom randomization algorithm or SQL server provides some functionality.

Answer 1

SELECT question_id
FROM questions 
ORDER BY NEWID()

gives you all the question ids in random order. store the list, and work your way down.

There isn’t a way to ensure the longest time between repeats without storing the order. You could make it another column on the same table…

EDIT: a variation on Mr Jacques’ idea:

how about a column named random? Shuffle the order like so:

update questions set random = (
    SELECT question_id
    FROM questions 
    ORDER BY NEWID()
)

then get each new question like this:

select * where random = 'prev_question_id';

Then you could shuffle the order at intervals, every night maybe.