Thanks for the keyword "parallelism deadlock" Grant. Lead to intra-query parallelism with more information on the internet.

As by your recommendation solved with a query hint.

We use SSIS but it is picky on metadata (so not really suited for excel files) and I found the error handling uneasy.

No user friendly error messages ( metavalidation failed...

Still wondering why something likely as importing a csv is still so hard using the sql server tools

Not sure what the question is.

Is it about a checklist what to do if there is an outage? Work it out with the team / management / stakeholders.

Is it about...

Not relevant, but why are you still developing for SQL Server 2008 RTM 11 years later ( RTM released in 2010 )?

Thanks Jeff. Now investigating sql plan guides for additional nudges.

Summer has arrived + 25 degrees Celsius

@TestContributions MONEY=NULL OUTPUT

you can find the value afterwards in @TestContributions

if you don't specify @TestContributions it will assume the default value NULL

Can you post the DDL of the tables so it...

No idea, have you tried auto adjust buffer size ? https://www.mssqltips.com/sqlservertip/4221/improving-data-flow-performance-with-ssis-autoadjustbuffersize-property/

I don't know of any official recommendation.

A practical reason could be that it is faster to relocate a 64 GB file than a 2TB one

There is an recommendation to have...

Apparently you are on service pack 2 https://sqlserverbuilds.blogspot.com/

Do you run it from commandline?

Alas, SQL Server's performance can suffer from encapsulating in UDF

https://www.brentozar.com/archive/2019/04/finding-froids-limits-testing-inlined-user-defined-functions/

https://www.mssqltips.com/sqlservertip/4689/understand-the-performance-behavior-of-sql-server-scalar-user-defined-functions/

Having a weird case today (still under investigation). Parallel clustered index scan slower dan serial index scan (forced by MAXDOP 1)

*edit* repartition streams spilled to tempdb 🙁 Repartition stream 1=0,2=5000000,3=0,4=5000000....

• This reply was modified 5 years, 2 months ago by Jo Pattyn.

It's really a talent. Some dumb questions are hidden gems. Why didn't we think of that before, does it still apply now ...

I haven't got problems when upgrading to 2019. It introduces adaptive joins which may be finicky in flip-flop scenarios (large & small dataset).