Sorry, forgot about the specs:

MS SQL Sever 2016 EE with 8 cores, 256 GByte of RAM.

The spell table has about 2 mn records which multiplies to 10bn considerung the time...

Hi Jeff,

maybe I can help you out with MicroStrategy a little bit. End users are business analysts with an high level of knowledge of the underlying data but not with...

• This reply was modified 3 years ago by  svenflock.

Hi all,

thanks for all your input. The tally function did the trick. Though, I had to split it up on a yearly basis to get it finished in about an...

• This reply was modified 3 years ago by  svenflock.
• This reply was modified 3 years ago by  svenflock.

Hi Jeff,

the translation worked :-). I will test it on the large data set right away. The date format is just Excel being Excel where I prepared the demo. I...

Hi,

thanks for your answers. What I mean is what Phil Parkin describes or what you see in my Excel screenshot.

• For each row there is a left and a right...

Hi,

sorry, my remote access to the server doesn't work since half a day. I couldn't run the I/O tests.

But Brian pointed it out: MSSQL joins with the whole table what...

Hi,

your comment is exactly my issue I have.

To confirm your assumptions:

• fact_BeH is partitioned by guelt_von (begin of time interval)
• fact_Datum is partitioned by ddt_id (calendar date)

First, let us stick...

Foreign keys do exist (see screenshot of ERD, I just removed them from the script).

As I am a novice to query plans: does the sqplan file help you to answer...

I am currently at work to give you the exact style of DDL. I will come back on Monday if it is ok with you.

Hi, thanks, I will look into that.

Last question: is this really the idea how to solve such issues, i.e. use a metadata tool to duplicate objects otherwise to be to...

Never heard of BIML before. I googled it just now. Can you recommend any free tool to generate the dtsx packages? command line would work, too.

Side remark: I do not understand why Microsoft is not able to build source and target operators which are able to handle parallel threads. It is bad coding design to...

Here is something I have in mind (just thinking)

I could extend the current hub structure

• Hash Key
• Business Key
• Load Date
• Source Sytem

with a new purely technical column running only...

Just to make it clear: staging brings only 50 million records, the lookup is done against 750+ million records. So partitioning the stage data only makes the 50 millions part...

I have done it first in pure SQL using a left join with where is null condition to keep only the new ones. For some reason, the query estimator was...