Informatica at run time creates different types of cache files depending on the transformations used in the mapping and configuration of the session.
Sometimes the cache file size can get so huge that the whole disk gets used up which causes fatal errors in other parallel running workflows and sessions. Huge cache files are also a performance bottleneck and the design should factor in steps to minimize the cache file size.
Following two steps can help in reducing the cache file size created by Informatica sessions:-
1. Joiner Cache and Aggregator cache can be reduced greatly if you provide them sorted inputs and check the sorted input flag in the config tab. Sorted input can be provided by passing data through a sorter transformation or sorting at database level in Source qualifier.
2. Lookup cache can be reduced if you only include the columns needed in the condition, return or output ports of the lookup i.e. by removing any other columns from the ports tab. Further improvement in lookup cache can be achieved by using sql query override to limit the rows by using a where clause to limit the overall lookup dataset.
Hope these two steps help you in preventing any unexpected failures at runtime due to space constraints