If you encounter a disc space issue in SAS CI360, MATables that have not been cleaned up can be the issue.
The “subject lists” that are downloaded from SAS CI360 are converted to MATables SAS data sets when necessary. Those data sets might be used for in-database processing (such as the customer’s data mart and/or CDM tables) or other SAS specific functionality such as stored processes.
Creating joins between SAS data sets and database tables in SAS is usually a slow process because all of the data from the database table needs to be downloaded into the SAS Platform Server to enable the join. SAS uses the database bulkload/fastload utility to load the MATables to the database to allow the joins to take place there, which is typically much faster.
Any MATables that are automatically uploaded to the data mart during the execution are removed after use when possible (depending on errors encountered, or database privileges). If the execution fails, an agent disconnection occurs, or the database or user does not have drop privileges, then the temporary tables in the database are not deleted/dropped.
If you encounter disc space issues, you can safely manually delete MATables older than a certain number of days (for example, older than 7 days). The following paragraphs provide information about the tables and why it's safe to delete them.
There is logic in the On-Premises agent that re-uses existing MATables for ad hoc counts requests. However, if those MATables are missing, they are regenerated as necessary. Note that workaround requires that the segment map is re-executed to regenerate those missing tables.
CI360 does not rely on pre-existing MATables for segments or link nodes. Instead, those are stored in CI360, which means that if those tables are removed, their dependencies are not broken. Unless you remove them while an execution using them is in progress.