Hi,
I have modelled fact table which store daily snapshot of closing balance of any account. A sample fact_table is:
time_id, account_id, balance
1, 1, 1000
1, 2, 5000
2, 1, 6000
As evident, balance is non-additive on time dimension.
If I understand correctly, for this daily snapshot + semi-additive to work properly, I have to populate all the accounts' remaining balance daily, even for those account which didn't change in balance amount. Because I am trying to closingperiod use approach mentioned in this post: http://forums.pentaho.com/showthread...itive-measures .
This approach increases fact table size considerably. Because on a given day only 10% accounts' balance change.
I am looking for some other approach which can work this new table population logic (where I store just changed balances everyday). This would help me reduce fact size by 90%.
Your help is appreciated. Thanks.
I have modelled fact table which store daily snapshot of closing balance of any account. A sample fact_table is:
time_id, account_id, balance
1, 1, 1000
1, 2, 5000
2, 1, 6000
As evident, balance is non-additive on time dimension.
If I understand correctly, for this daily snapshot + semi-additive to work properly, I have to populate all the accounts' remaining balance daily, even for those account which didn't change in balance amount. Because I am trying to closingperiod use approach mentioned in this post: http://forums.pentaho.com/showthread...itive-measures .
This approach increases fact table size considerably. Because on a given day only 10% accounts' balance change.
I am looking for some other approach which can work this new table population logic (where I store just changed balances everyday). This would help me reduce fact size by 90%.
Your help is appreciated. Thanks.
