Hi,
Starting point for Information Lifecycle Management are the business users' requirements for storing the data for a certain time period. This will depend on the type of information. So it will be wise to categorize the data and define data retention periods per category.
The technical solution is partitioning. Logical partitioning on a Time Characteristic like Calendar Year or Fiscal Year is very useful. This can be combined with table partitioning on Calendar Month or Fiscal Period depending on your database.
This will give you a scalable solution and easy to handle from a data management point-of-view. You can for example store certain logical partitions on Near-line Storage (NLS) if you have that in place. Also dropping the oldest partition is quite easy to manage.
For reporting a MultiProvider can be used as a virtual layer between persistent InfoProviders and BEx Queries.
A note on data volumes per InfoProvider (PartProvider): it does not make sense to use logical partitioning with very small data sets. It will depend on your database but be aware that data volumes of 200 or 300 million records per InfoProvider are possible for databases like Oracle and DB2. I would suggest to have a lower limit per InfoProvider and also take the advantages of "yearly" partitions into account. But if you have less than e.g. 10 million records, I would not recommend using logical partitioning.
I would recommend to have use SPO (Semantically Partitioned Object) to facilitate the logical partitioning. Please have a look at my blog Pattern-based Partitioning using the SPO BAdI - Part 1: Introduction for more information and an idea how you can introduce partitioning patterns using the SPO BAdI.
Best regards,
Sander