This particular article analysis some of the tips and techniques that we secured inside my webinar Overall performance Strategies for Electricity BI Analysis Models . There are more resources and methods, in addition to expanded causes and you can sources regarding speech glides.
My analogy investigation because of it chat is a customized Work dining table away from Sales team. So it real world data desk is inspired by a consumer and that’s sanitized having privacy. The fresh new analogy enjoys more 382,one hundred thousand rows and this is not a giant desk for Electricity BI. When the info is loaded for the Stamina BI, the fresh new kept file size towards the computer balloons to over 500MB. Into the thoughts this info lay occupies more than 1GB.
- Formula edits was sluggish
- Slicers towards the accounts slowed down
- Refresh of information got extended to help you stream and you may processes the job desk
What makes that it table awkward toward columnar study shop is actually partly the fresh new string articles, thirteen of them was over 2 hundred letters much time. One career format are a likely signal you to definitely text thinking try joined of the typing and this have numerous more beliefs, probably in just about every non-empty field.
All suggestion these might not affect the challenge. For more information on each idea, delight review the latest slides away from my personal demonstration otherwise look at the submitted webinar right here.
When it comes to tables that have 10’s from millions of rows, the brand new 1M-line partitions or “Chunking” can get interfere with efficient compression pricing, delight check this out to learn more:
Perhaps you have realized from the data stream timing table lower than, and you may in the picture of comparison file brands, just after some of the higher cardinality (most unique) highest fields were eliminated, the fresh quality dropped additionally the studies loaded faster.