Question

I want to store data in MySQL and query it based on the current day. I want to know what is the best practice to do so.

I want to store data totals for each day, so queries total data will be quick. I thought about modeling my table as follows:

TotalsByCountry
- Year
- Month
- Day
- countryId
- totalNumber

When I query the totals for a specific day and for specific country, I will query the table based on 4 columns, the Year, Month, Day and countryId.

I wanted to know if this is a good practice, or there is a better way to do so, like using one columns for data that holds the month, day and year, and query only two columns, the datetime columns and the coutryId.

need you help in choosing the right way to model the table. I also want to make another table that store totals based on gender, so take that into consideration too.

The data will need to be accessed frequently, maybe in real time because I want to show the data changes in real time. I will be developing the web app in asp.net and probably use web sockets to create constant connection that will update the data on the user in real time. So when data changes, it will be reflected on the user webpage in real-time. That’s why I need a table modeling that will be ready for many queries. I will use caching for a few seconds so it want stress the db too much.

I hope I provided enough information, if not, please comment and I will reply.

Answer 1

Having three separate columns to store each individual element of a date (year/month/day) will add unnecessary overhead to your database in terms of insert performance and disk space.

What you will want to do is simply have a single DATETIME column to store the date and time, and have a composite index set up on (countryId, datetime_col).

Even if you wanted to query all rows based on a specific day or month, MySQL will still be able to utilize indexes on the DATETIME field, provided that you are writing your queries in the right way and making sure to never to wrap the DATETIME column within a function when you perform your conditional check.

Here is how you can write your query so that it will still be able to utilize indexes:

-- Get the sum of totalNumber of all rows based on current day
-- where countryId = 1

SELECT SUM(totalNumber) AS totalsum
FROM   tbl
WHERE  countryId = 1 AND
       datetime_col >= CAST(CURDATE() AS DATETIME) AND
       datetime_col <  CAST(CURDATE() + INTERVAL 1 DAY AS DATETIME)

By making the comparison on the bare DATETIME column, the query remains sargable(i.e. able to utilize index range scans) and MySQL will be able to use indexes to quickly look up rows.

On the other hand, if you were to try to wrap the DATETIME column within a function to make the comparison:

-- Get the sum of totalNumber of all rows based on current day
-- where countryId = 1

SELECT SUM(totalNumber) AS totalsum
FROM   tbl
WHERE  countryId = 1 AND
       DATE(datetime_col) = CURDATE()

…It would be quite inefficient because the DATE() function that wraps the column effectively renders the query as non-sargable, and any kind of index you have set up containing the DATETIME column will not be utilized.

You can also efficiently query for the total sum of all rows in the current month:

-- Get the sum of totalNumber of all rows based on current month
-- where countryId = 1

SELECT SUM(totalNumber) AS monthsum
FROM   tbl
WHERE  countryId = 1 AND
       datetime_col >= CAST(CONCAT(YEAR(NOW()), '-', MONTH(NOW()), '-01') AS DATETIME) AND
       datetime_col <  CAST(CONCAT(YEAR(NOW()), '-', MONTH(NOW()), '-01') AS DATETIME) + INTERVAL 1 MONTH

And within the current year:

-- Get the sum of totalNumber of all rows based on current year
-- where countryId = 1

SELECT SUM(totalNumber) AS yearsum
FROM   tbl
WHERE  countryId = 1 AND
       datetime_col >= CAST(CONCAT(YEAR(NOW()), '-01-01') AS DATETIME) AND
       datetime_col <  CAST(CONCAT(YEAR(NOW()), '-01-01') AS DATETIME) + INTERVAL 1 YEAR