Archives of the TeradataForum

Message Posted: Wed, 02 Dec 2009 @ 16:21:22 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

Anne: High level, there are two approaches, 1 near term, 1 long term.

- Near term: Replication is performed using ARCMAIN "archive" and then "copy" (not "restore"). The advantage of COPY is that it always lays down and exact replication of the table and of the data. The process is driven by a table containing the ETL jobs and the tables that are touched. When the ETL job is complete, the scheduler runs a set of objects that read the table of jobs and runs the ARC process for the tables associated with the ETL job id. This is near term, because it can ramp up rapidly.

-Long Term: Replication is by creating "dual stream" ETL processes. Whatever runs against the prod server, is simultaneously run against the dr server. The same input files are used by the prod and dr server loading processes. Long term is because takes longer to implement.

Very high level, but is the core process.

	<Prev	Next>		<<First	<Prev	Next>	Last>>