NIM is faster than other provisioning tools because it isn't built on a traditional database model. Instead, NIM loads the current dataset into RAM at the beginning of the session, and keeps it there as long as NIM is running. This area of RAM is the vault.

Since different external Systems have mutually incompatible data formats—sometimes even totally proprietary data formats—NIM transforms all collected data into a universal, internally consistent format during Collection. This transformation is defined in the systems' underlying Connectors. The resulting data is stored in the vault. This data, as interpreted through your configured Data models and Inter-system relations, becomes the basis for all Filters and ultimately all provisioning operations performed by NIM's Soll-ist engine.

NIM both reads and writes to the vault, but at different times. During Collection, NIM reads the latest data from the external system and writes it into the vault. (If the vault already contains data for that system, the old data in the vault is overwritten.) Oppositely, when your Jobs are executed, NIM calculates the soll differentials and writes the results into the external systems. Thus, the vault is an "intermediary" between NIM and your external systems. Together, collection plus job execution comprises the soll-ist loop. See Soll-ist engine.

Although NIM only operates on data currently in the vault, it does retain previously collected data on your hard disk. These historical collections are stored on a per-system basis. In this state, this data is inactive and not available to your filters.

Note that there is an important distinction between collecting and loading data. They are not identical operations. Collecting data necessarily loads the collected data into the vault, but loading data does not necessarily collect new data. It is possible to Load a past system collection without performing a new collection. To view all past system collections that are available to be loaded, View past system collections. This is a relatively uncommon procedure. Most of the time, you will not need to load previous collections back into the vault.

To manually update a system's data in the vault, Collect and load a system.


Most of the time, the vault operates invisibly, behind the scenes. The vault has no dedicated screen in NIM. To "see" the current vault contents, simply Create a filter. The tables available in the filter are the current contents of the vault.