Near Line Storage ( NLS ) Archival with SAP BW - Part 1

[ankush_hallan2]

Part 1 of this blog series will cover some important aspects of NLS archival technology along with SAP BW .

Some important points to keep in mind before starting a data archival project are :

1) Choosing a NLS vendor . There are many SAP certified NLS vendors and a vendor should be chosen based on

requirement and product offered .

2) Correct sizing of the NLS box is important so that current and future data archival needs can be met.

3) Data to be archived : Archived data is always online for reporting while using NLS , but the reporting speed is

not comparable to reports being run on BWA or HANA appliances . Only data which is rarely accessed should be

put on NLS .

The next step is to determine which InfoCubes and the data volume that should be archived .

Based on BW statistics this decision can be made along with inputs from the business users . InfoCubes which

are big in size and infrequently used  are the best candidates for data archival .

Below is quick chart of the logic that can be  used to determine how a particular InfoCube should be archived

NLS archival is done using time slices and it is important that InfoCubes have time characteristics which can

be used for data archival .

InfoCubes which are refreshed on a daily basis ( Full load ) need not be archived . If SAP ECC system or

source system is also undergoing archival then only live data post archival in the SAP ECC or source system

will get loaded as part of the daily full loads to such InfoCubes and data which has been archived will not be

available in the SAP BW system for reporting .

InfoCubes which are delta enabled and delta loads don't bring data for time slices which have already been

archived are the best candidates for NLS archival . The reason here is that if you archive data from a cube for

a particular time slice and if the daily delta brings in data for the same time slice the data loads will fail .

If your daily delta loads to InfoCubes bring in data for time slices which have already been archived it's better to

create copy cubes and move the data to copy cubes for those particular time slices and archive those copy

cubes instead of the original cubes. Post validation , data can be deleted from original InfoCubes . Existing

multi providers and reports will have to be enhanced to include the new copy InfoCubes .

Every business requirement,landscape is different and there are multiple ways data archival can be done using NLS .

Above is one approach that can be used .

The second part of this series will try to cover some more aspects of NLS archival technology

Cheers !