In Addition to the response by Jürgen L.
My approach would be like this .
1. Get the complete end to end business process that is currently running in your system (ex: SD,
FI, etc) . there should always a start and end. within this identify all those transactions like (sales orders, PO, FI docs etc) and also note the sequence in which they are getting created in the system.
2. once you have this , you can identify relevant archive objects for each of the business transaction.
for archiving we follow reverse sequence i.e the document that was last posted and completed is archived first. ex: check in SARA SD_VBAK (network graphic icon)
3. research OSS notes with each archive object name : you will get more details as what the archive or preprocess program checks before archiving .
if you identify there are additional checks that you need , then use BADI's to modify the standard programs. ex: 456129
- FAQ: Archiving in purchasing
4. note that for any document to be archived , a basic pre-requisite is that it should be business complete, closed and residence time reached.
5. consider any custom fields,custom tables, reports, interfaces also for impact analysis.
many times its hard to convince business for archiving on the above .
6. the complexity is different for each archive object. the more complex , the higher analysis required.