When SQL Server receives a new query, it attempts determine the best possible plan for resolving that query. It considers a number of different factors as it analyzes the query and maps out a way in which to retrieve the information requested.
Whether or not the query optimizer deems an index to be useful in resolving a query largely depends on the information contained in the statistics for that index.
If the statistics are outdated and do not accurately represent the distribution of values within the table, the query optimizer may not produce an optimal plan for resolving the query. Misleading statistics may result in the optimizer not using an index when it should, or using an index when it would be more efficient to scan the table.
That's why it is crucial that the statistics for an index be updated regularly. How often? Well, that, of course, is going to depend on the data being stored, the frequency of updates, inserts, and deletes to the table, etc. It could be nightly; it could monthly. It just depends.
So, how can you tell when the statistics where last updated for an index?
The following query demonstrates this. It makes use of the sys.indexes and sys.tables catalog views, along with the STATS_DATE() function, to retrieve the date that each index was last updated for every user table in the current database.
SELECT
t.name AS Table_Name
,i.name AS Index_Name
,i.type_desc AS Index_Type
,STATS_DATE(i.object_id,i.index_id) AS Date_Updated
FROM
sys.indexes i JOIN
sys.tables t ON t.object_id = i.object_id
WHERE
i.type > 0
ORDER BY
t.name ASC
,i.type_desc ASC
,i.name ASC
This little query can be useful in troubleshooting and diagnosing performance-related issues. Sometimes, it's as simple as outdated statistics.
Cheers!
Joe
