Today I learned that apart from archiving old data to a file system, you have an option to store the data on a separate database, so that BW queries on old data can still be used for reporting.
This is very useful when using BW on HANA.
SAP BW powered by HANA, uses a multitemparature concept when deciding which data must be stored in memory and which data to push out of memory.
The data on which reporting is done on a daily basis is HOT data.
The data on which reporting is not daily or when you don't do any reporting at all (ex PSA tables and write optimised dso tables containing corporate memory) is WARM data. Warm data is offloaded from memory, when there is memory problem, to disk. This still ties the data to the database and is included with database backups and restore. SAP HANA uses a separate engine to carry out the dynamic tiering concept to offload such data.
The data that does not change (historical data) can be shifted to a near line storage. For BWoH, SAP IQ is the NLS of choice. SAP IQ can operate on columnar store paradigm as is the case with HANA. The hardware requirements of SAP IQ are cheaper, therefore it is a good choice to archive data out of HANA. And most importantly SAP BW has a near line storage interface to help run queries seamlessly on SAP IQ.