In the target properties you can set number of rows after which a commit has to take place. In the flow properties you can set restart from last commit. Suppose I join three source tables to fill a target table. Commit after 1000 rows. The job abends after 2090 inserts/updates.
How does restarting work then? When I run the job again, will it first make for a second time the complete join of the three source tables? So, here no processing gain? And will it then start insert/updating starting from record 2001?
Where does the tool keep this information, this restarting point?
Are there essential things I should keep in mind when working with commit/restart mechanisme?
Thanks RonThis message has been edited. Last edited by: FP Mod Chuck,
That is essentially correct. The only performance gain would be that the first 2000 records need not be loaded again.
Use of restart from the last commit point should be reserved for static (or at least append only) data sources such as flat files or database tables that will not have rows updated or deleted. Also for most RDBMS the order that rows are returned is undefined so the answer set should be sorted to ensure it appears in the same order for subsequent runs.
N/A
April 06, 2018, 04:08 AM
Ronibi
Hai Clif
thanks for your answer. We are working with tables. Do you say that we should reserve the restart/commit functionality when we do merely inserts on the target table? Or do you mean that between an abnormal end and restart of the flow the source tables may not be changed? Which I understand.
regards RonThis message has been edited. Last edited by: Ronibi,