Updating Monthly Actuals - Better Approach?

[Former Member]

Does anyone have any experience or comments on these two approaches to the monthly process of loading actuals?

1. Clear all fact records for existing month. Load the file data as-is.

2. Keep existing fact records. Load the variance between current file and previous file (stored in a staging database)

Wouldn't the second approach be ideal since the process would only get faster towards the end of month? It would also generate fewer records. Generally, the file will produce about 90K records each month.

[sorin_radulescu]

1. This process clear data - means the BPC is adding the value into the system with negative sign and you will 0 for values of that month

Importing data after is adding the valuefrom your file into database

2. This process is doing something similar with first:

first is checking wwhat records are you importing into database for dimensions used by your application into work status.

Usually Category, Time, Entity and after that is adding the diference into datbase only for tuples Category, Time, Entity from File.

But even in this case is doing first a clear for that selected tuples and after that is adding the new values into system

In my opinion I will say that first process is better and easier to follow. Depending by how big is the import file the second process is an option.

For big files I suggest first approach for small file probably is better the second but even in that case I suggest to use the first one because it is easier to understand where is the problem if you have some bad values into database.

Regards

Sorin Radulescu