Is it possible to write a Microsoft SQL query that will group by a datetime
data-type but ignoring the time part such as the hour and minute?
To group by date part, use the GROUP BY clause and the EXTRACT() function. Pass EXTRACT() the date parts to isolate.
When I need an answer in units other than days, I will use DateDiff. Note that this only applies to the old DATETIME type, not to DATE , TIME , or DATETIME2 . Also, the value returned will be another DATETIME , so you'll need to cast it to get the human-readable number of days between dates.
Using Group By and Order By Together When combining the Group By and Order By clauses, it is important to bear in mind that, in terms of placement within a SELECT statement: The GROUP BY clause is placed after the WHERE clause. The GROUP BY clause is placed before the ORDER BY clause.
To compare dates without the time part, don't use the DATEDIFF() or any other function on both sides of the comparison in a WHERE clause. Instead, put CAST() on the parameter and compare using >= and < operators.
If you are on SQL Server 2008 this is simple.
GROUP BY CAST(YourCol AS Date)
For previous versions you can use
GROUP BY DATEDIFF(DAY, 0, YourCol)
However neither of these will be able to leverage the fact that an index on YourCol
that is ordered by datetime
will also be ordered by date
and thus use a stream aggregate without a sort operation.
On SQL Server 2008+ you might consider indexing (date,time)
rather than datetime
to facilitate this type of query.
Either by simply storing it as two separate components and possibly providing a calculated column that recombines the parts (a datetime2
is stored the same as a date
and a time
so this won't consume any more space except for if the additional column pushes the NULL_BITMAP
onto a new byte.).
CREATE TABLE T
(
YourDateCol date,
YourTimeCol time,
YourDateTimeCol AS DATEADD(day,
DATEDIFF(DAY,0,YourDateCol),
CAST(YourTimeCol AS DATETIME2(7)))
/*Other Columns*/
)
Or alternatively you could store it combined in the base table and have the indexes use calculated columns that split it out.
An example of this approach
CREATE TABLE T
(
DT DATETIME2,
D AS CAST(DT AS DATE),
T AS CAST(DT AS TIME)
)
CREATE INDEX IX1 ON T(DT);
SELECT COUNT(*),
CAST(DT AS DATE)
FROM T
GROUP BY CAST(DT AS DATE)
CREATE INDEX IX2 ON T(D,T);
SELECT COUNT(*),
CAST(DT AS DATE)
FROM T
GROUP BY CAST(DT AS DATE)
DROP TABLE T
If you have a large table and want grouping to happen fast, you cannot rely grouping on an expression.
Create an additional datetime column on your table and have it filled by a trigger that calculates the base date from you "real" date:
UPDATE
MyTable
SET
EffectiveDate = DATEADD(DAY, 0, DATEDIFF(DAY, 0, t.RecordDate))
FROM
MyTable t
INNER JOIN inserted i ON i.RowID = t.RowID
Then put an index on EffectiveDate
and grouping (and searching) will be a lot faster.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With