26,075 reputation
656103
bio website
location United Kingdom
age
visits member for 2 years, 6 months
seen Apr 3 at 19:03

Email: stackoverflowmartinsmith@gmail.com


Jan
22
comment Why does subquery use parallelism and join doesn't?
@AaronBertrand - But then there is the question of why there are much more way off for the join plan than the sub query when they are both based on the same thing...
Jan
22
comment Why does subquery use parallelism and join doesn't?
And totally different plan shapes as well with the serial plan having hash and sort spills due to the one row estimate and insufficient memory grant.
Jan
22
comment Why does subquery use parallelism and join doesn't?
The join version is only costed at 0.96327 this is probably below the cost threshold for parallelism for a parallel plan to be considered. Significantly undercosted as the estimated number of rows is much lower than the actual.
Jan
22
comment Why does subquery use parallelism and join doesn't?
@ChrisL People generally copy up to sites like pastebin and just provide the link.
Jan
22
comment Why does subquery use parallelism and join doesn't?
Can you post the XML version of both actual execution plans?
Jan
19
revised How do I query for fields that have AT LEAST a given range?
added 70 characters in body
Jan
19
answered How do I query for fields that have AT LEAST a given range?
Jan
19
comment How do I query for fields that have AT LEAST a given range?
What if the value has gaps? Is a range of 4 a contiguous series of 4 values? Or just the difference between MIN and MAX for that key?
Jan
19
revised Is it possible to view LRU-K values in SQL Server?
Forgot definition of table T
Jan
19
comment Need advice on abnormal login auditing
Yes looks like a dictionary attack against those accounts. 61.138.99.12 is a Chinese IP address BTW.
Jan
18
revised Why is SQL Server using a clustered index scan for a self referencing FK cascade delete
added 599 characters in body
Jan
18
comment Capture Complex Actual Execution Plans
If you have all the individual XML plans isn't it quite trivial to consolidate them into one XML document?
Jan
17
comment Why is SQL Server using a clustered index scan for a self referencing FK cascade delete
@Vaccano - Which proves the point about estimated costs! If the scan was only 20% of the query cost the maximum improvement that could be expected from removing it would be 12 seconds.
Jan
17
revised Why is SQL Server using a clustered index scan for a self referencing FK cascade delete
added 338 characters in body
Jan
17
comment Why is SQL Server using a clustered index scan for a self referencing FK cascade delete
@AaronBertrand - Yes. Looks like it just uses the same containment/inclusion assumptions as it would for a normal query against the table. i.e. it costs as though it "expects" the delete to fail.
Jan
17
revised Why is SQL Server using a clustered index scan for a self referencing FK cascade delete
added 441 characters in body
Jan
17
answered Why is SQL Server using a clustered index scan for a self referencing FK cascade delete
Jan
16
comment When changing the size of a nvarchar column, do I need to drop the unique index? And will the table get locked when recreating the index?
@AaronBertrand - I suppose it's just a missing optimisation for the cases where it isn't required. Books online indicated that this was needed in a few cases where the product did not actually require it as well until this update
Jan
16
revised When changing the size of a nvarchar column, do I need to drop the unique index? And will the table get locked when recreating the index?
added 269 characters in body
Jan
16
answered When changing the size of a nvarchar column, do I need to drop the unique index? And will the table get locked when recreating the index?