Take a look at sys.dm_db_index_usage_stats, it should give you what you want.

With the exception of random but evenly distributed indexes, I wouldn't use logical fragmentation to determine if an index needs to be rebuilt.  In the case mentioned, you actually...

Wrap the value in a: FLOOR, CEILING or ROUND(,0) function, depending on how you want to handle the decimal part of the value.

There's not a lot of data left in the file, so I'm rather surprised it would take that long to shrink it, since SQL wouldn't need to move that much...

The first command might not shrink the file at all, because of the "TRUNCATEONLY" option.

The second command should work, but shrinks can take a long time.

For best performance, you want to stop using NULL so you don't have to do ISNULL() as part of the WHERE.  Yes, you will have to go back...

Once you get rid of the LOWER on the vehicle type, then the clustering key could be ( VehicleType, ValidFrom, ValidUntil ).

For best performance, you want to stop using NULL so you don't have to do ISNULL() as part of the WHERE.  Yes, you will have to go back and change...

SELECT LEFT(devicename, month_in_date - 2) AS device, MAX(CAST(devicedate AS datetime)) AS max_device_date
FROM #test1
CROSS APPLY (
    SELECT PATINDEX('%[0-9]/%', devicename) AS month_in_date_prelim
) AS ca1
CROSS APPLY (
...

I think CHARINDEXes might be somewhat more efficient here:

;WITH data AS (
    SELECT rl_event_id = 'AB123456_BlahBlah_BESTSELLER_blahBlah'
)
SELECT rl_event_id, SUBSTRING(rl_event_id, pos_of_second_underscore + 1, pos_of_third_underscore - pos_of_second_underscore...

The db that table was in already had space allocated but that had not been used yet.  Therefore, SQL can allocate that space to a table without having to get...

RAM doesn't look too short there, but personally I'd add more if you could.  RAM is the cheapest overall performance boost you can get.  And bump up tempdb size if...

dup, removed

Update T --<<-- *must* use T here (the alias and NOT the table name) to be safe
SET ACTIVE=CASE WHEN S.user IS NULL THEN 'N' ELSE 'Y' END
FROM...

(1) Cluster the "FF" table on ( FileDate, ISIN ) (if it's not already clustered that way).

(2) Some of the JOIN conditions are out of place.  Change this code:

            
August 22, 2023 at 2:41 pm
            
#4268207