Reputation
37,116
Next tag badge:
542/400 score
74/80 answers
Badges
8 97 156
Newest
 Enlightened
Impact
~1.4m people reached

Nov
4
comment Why does LEN() function badly underestimate cardinality in SQL Server 2014?
LEN itself can be constant folded. SELECT COUNT(*) FROM #customers WHERE LEN('FOO') = 6 doesn't access the table at all.
Oct
23
awarded  Good Answer
Oct
23
comment Performance inner join vs left join involving parallelism
would need to see the plans so unless you can provide a repro with dummy data?
Oct
22
comment Performance inner join vs left join involving parallelism
Also is the actual query parameterised?
Oct
22
comment Performance inner join vs left join involving parallelism
So I presume that the faster query is using different join algorithms (eg hash vs nested loops) and the difference isn't just parallel vs serial then. You should try updating statistics on the associated tables.
Oct
22
comment Performance inner join vs left join involving parallelism
@Shnugo A left join can't produce fewer rows than an inner join if that's all that changed.
Oct
22
comment Performance inner join vs left join involving parallelism
You can find out the DOP by looking at the properties of the root node in the actual execution plan. The cost threshold for parellelism by looking at sys.configurations and the estimated cost of the inner join plan from selecting the root node in either the estimated or actual plan and looking at the properties. And whilst you're there look and see if there is a reason for non parallel plan.
Oct
22
comment Performance inner join vs left join involving parallelism
What is the estimated cost of the inner join plan? What is your threshold for parallelism set to? Also what DOP do you actually get in the parallel plan? That is quite a speed up!
Oct
22
comment How can I rebuild an index on table with 100 million rows without a wait time out?
Run the script through a new query window in SSMS. Not through the visual designer.
Oct
21
comment Create index with a commentary
Your bow tie question has been answered....
Oct
21
revised SSIS Merge component performing a UNION instead of merging
Components are control flow
Oct
21
comment Estimated row size and estimated data size is incorrect
I suspect that is calculated from actual rows * estimated data size.
Oct
21
comment Estimated row size and estimated data size is incorrect
No there's no way of adjusting it apart from changing the column size either at source or with a cast. The snippet of execution plan you have shown just has it going straight into an insert to a heap anyway so I don't think that the poor estimate would have any particular effects except if there are other indexes on that heap including those columns.
Oct
21
revised Estimated row size and estimated data size is incorrect
added 58 characters in body
Oct
21
answered Estimated row size and estimated data size is incorrect
Oct
20
comment How to estimate rate of Trans Log usage
sys.dm_io_virtual_file_stats will show the amount of data written to the log.
Oct
20
comment prevent clustered index insert operator on unqualified indexed view
Not in a position to test but does adding option (recompile) help?
Oct
15
awarded  Good Question
Oct
13
revised Query 100x slower in SQL 2014, Row Count Spool row estimate the culprit?
rolled back to a previous revision
Oct
13
comment Query 100x slower in SQL 2014, Row Count Spool row estimate the culprit?
@GeoffPatterson thanks for the extra info...