Forum Discussion
Size Of Publishing Compromise Implementation Scenario
Hi Viera,
How many columns do you have in the 13million row table you're pulling into Power BI Desktop and how many total tables are in your model?
Is this in a dimensional model, otherwise known as an EDW, a star schema, or a snowflake schema, or was the table made by joining many other tables onto the 13 million row table? A very wide table can take up a lot of space in memory depending on what sort of data is being compressed, particularly if there are numerous columns with unique values.
Or, are you just pulling in the SQL Server tables directly, as in are these operational tables?
I'm asking this as the nature of the data may indicate the model could take up less size if shaped in a different way.
Best of luck either way,
Alex
- viera0010 years agoHelper II
HI AlexBarbeau,
Q: How many columns do you have in the 13million row table you're pulling into Power BI Desktop and how many total tables are in your model?
A: My table has almost 50 columns, total tables only 5.
Q: Is this in a dimensional model, otherwise known as an EDW, a star schema, or a snowflake schema, or was the table made by joining many other tables onto the 13 million row table?
A: Is a plain table from a transactional system. No dimensional model.
Q: A very wide table can take up a lot of space in memory depending on what sort of data is being compressed, particularly if there are numerous columns with unique values.
A: Yes, its true, only the table with a couple of relationships to other small tables is consuming huge space.
Q: Or, are you just pulling in the SQL Server tables directly, as in are these operational tables?
A: Only pulling 1 table, that has lots of columns.
Q:I'm asking this as the nature of the data may indicate the model could take up less size if shaped in a different way.
A: Sure, I understand this is not particular optimized. But in the future this table will grow very much. I'm thinking in a start model, grouping values by a couple of dimensions. But even with this grouping, I feel the facts table will still be huge.
As a conclusion, I think i will move to a SSAS tabular, as working only with the file is not really an option at this point.
Regards.