Persist Node

The persist node stores the state of the database at a particular point in a stream; this speeds up subsequent runs of the stream.

If, for example, a database had been extracted from a remote source followed by the discretisation of several fields and the selection of the best features, you could persist the stream at this point in order to save the state of the database to memory (where available). If a discovery node and a decision tree node were then linked to the persist node and the stream run several times, the database would be loaded from the persisted state in memory rather than having to retrieve it from the original location and reapply the pre-processing nodes. When the database state is initially being persisted, the node first attempts to save the database state to physical memory; if there is insufficient physical memory available it will try to save it in virtual memory (using the Windows paging file). Should both of these activities fail, the stream will continue to run but the database will not have been persisted and future runs will have to execute the entire stream.

Options

Full details of the options available for the persist node can be found on the Persist Node options page.

Notes

The state of the persist node is shown using two different icons on the workspace:

 

	Persist node is active and the data is available.
	Persist node is inactive. The data will be created when the stream is next run.