Housekeeping Part 2 - System Tables

The housekeeping tasks in this section are a bit more technical and can’t be run as part of process chains. In order to run these tasks regularly you will need to set up program variants & background jobs to run them for you.

There are too many individual tasks to detail in this blog posting, so I will just mention a couple to get you started. Have a look at the SAP note I mentioned in the introduction for a complete list of all the system tables which can grow large, and instructions about how to trim each of them.

Archive request tables using BWREQARCH in SARA

Each time a dataload is executed in BW, entries are written to several system log tables. These tables specifically relate to BW dataloads, and only exist in BW systems. Eventually these tables will grow very large, so it’s a good idea to trim them. You can’t just delete old data from them since the system needs access to the data in order to function properly, so SAP have provided an archiving solution via the BWREQARCH object in transaction SARA which will allow you to archive off old log data onto the filesystem and then delete it from the database. This reduces the size of the database and increases system performance, but unfortunately does mean that you are taking up space on the filesystem when you store the archived filed. However, storage is cheap so this should not be too much of an issue.

Delete old system log entries (just like in ERP)

All other non-BW-specific system activities are logged in the system log, just as in ERP systems. This log can get very large and so you should trim it, just as you would on an ERP system.

BW Statistics

BW is often configured out of the box to log detailed performance statistics data so that this can be loaded into the BW technical content cubes (also known as BW statistics cubes). The tables which hold this statistics data can get very large, especially if you are logging the data but not then loading it into the cubes. You should configure the collection of BW statistics data to suit your needs, and also delete unwanted data from the statistics tables.

Guest author: Mark Wheaton