Vectorial storage: how does it work?

[nicolas_martin]

Hello,

I have a table with about 50'000'000 prices.

This tables contains the following columns:

- id (a unique ID)

- product_id

- country_id

- price_date

- price_excluding_VAT

This table weight about 500 Mo in a QVD.

I have a table with the possible VAT by day and by country:

- country_id

- VAT_date

- VAT (in %)

This table is about 1 Mo.

If I make a load in 2 tables, I have a $Syn key, and it's harder for me to do my expressions, but my QVW file is about 500 Mo.

If I make a LEFT JOIN between the 2 tables using the columns "country_id" and "VAT_date / price_date", this adds a field "VAT" in my price table.

I thought that it would not weigh down my QVW file, because it's often the same VAT from a day to another, and QV uses a vectorial storage.

Unfortunatly, my QVW file is now 800 Mo!

(and my join is correct, because I don't have any duplicated "id" after my join)

I think I missed something in the "vectorial storage" behavior.

Can anybody explain to me why I have such a big file?

[nicolas_martin]

Indeed, I used a "High" compression.

When I disable the compression, my files are approx. the same size: 93 MB.

So, the difference comes from the compression...

Adding a simple value can disturb the compression method depending on "how" the field is added.

Thank you for your answer!