Question

I have a table with bookings:

• id (int)
• quantity (tinyint)
• date_start (timestamp)
• date_end (timestamp)

Bookings can overlap, so there can for example be 20 bookings at the same time. Now I want to do two queries:

1) Retrieve the maximum sum of quantities within a given period.

2) Retrieve the minimum sum of quantities within a given period.

I have no idea how to add the quantities of ONLY the overlapping bookings together, and not of non-overlapping bookings. Any help would be greatly appreciated.

edit: i drew a picture for clarity: https://i.stack.imgur.com/I9nvq.png

Answer 1

1. Construct a list of all booking boundaries (i.e. start and end dates) which occur in the desired period:

     SELECT date_start AS boundary
     FROM   bookings
     WHERE  date_start BETWEEN @start AND @end
   UNION
     SELECT date_end
     FROM   bookings
     WHERE  date_end BETWEEN @start AND @end
   

2. Add to that the boundary which occurs immediately before the desired period:

   -- [ from part 1 above ]
   UNION
     SELECT MAX(boundary)
     FROM (
       SELECT MAX(date_start) AS boundary
       FROM   bookings
       WHERE  date_start <= @start
     UNION ALL
       SELECT MAX(date_end)
       FROM   bookings
       WHERE  date_end <= @end
     ) t
   

3. Make an outer join between this result and the bookings table, keeping all of the boundaries but only including a booking if it contributes to the number of simultaneous people after the boundary:

   FROM bookings RIGHT JOIN (
     -- [ from part 2 above ]
   ) t ON date_start <= boundary AND boundary < date_end
   

4. Sum the number of people at each boundary:

   SELECT IFNULL(SUM(quantity),0) AS simultaneous_people
   -- [ from part 3 above ]
   GROUP BY boundary
   

5. Find the maximum and minimum:

   SELECT MIN(simultaneous_people),
          MAX(simultaneous_people)
   FROM (
     -- [ from part 4 above ]
   ) t
   

Putting it all together:

SELECT MIN(simultaneous_people),
       MAX(simultaneous_people)
FROM (
  SELECT IFNULL(SUM(quantity),0) AS simultaneous_people
  FROM   bookings RIGHT JOIN (
    SELECT date_start AS boundary
    FROM   bookings
    WHERE  date_start BETWEEN @start AND @end
  UNION
    SELECT date_end
    FROM   bookings
    WHERE  date_end BETWEEN @start AND @end
  UNION
    SELECT MAX(boundary)
    FROM (
      SELECT MAX(date_start) AS boundary
      FROM   bookings
      WHERE  date_start <= @start
    UNION ALL
      SELECT MAX(date_end)
      FROM   bookings
      WHERE  date_end <= @end
    ) t
  ) t ON date_start <= boundary AND boundary < date_end
  GROUP BY boundary
) t

See it on sqlfiddle.