Archives of the TeradataForum

Message Posted: Thu, 25 Oct 2001 @ 16:10:54 GMT

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

We use several media and means for archiving data in our Production and Development environments.

For our Production environment...

- Monthly, we perform a system archive using two 6400 and two 6466 silos.

- Of course, a 6400 silo contains 10 DLT drives and a 6466 contains 12 9840 drives.

- We archive all data and DDL at that time.

- Each tape drive is connected to a different node, so we get excellent I/O rates.

- Part of this archive process includes two cluster archives to the 6466 silos for two particularly large tables that total about 1.5 terabytes.

- We can archive about 16.8 terabytes in about 14 to 17.5 hours depending on tape or drive problems.

- The analyze phase can run about 12.5 hours.

- The remainder of our archives are performed on several MVS mainframes using Magstar tapes.

- I don't have handy figures for these, but the weekly archives can take around 14 hours.

- Weekly we perform archives of more volatile data.

- Nightly archives are performed for even more volatile data.

- Finally, we perform checkpoints daily on our most volatile data every other hour.

- The journals are archived nightly also.

For our Development environments...

- We perform only a monthly system archive.

- This archive is done to Magstar tapes on an MVS mainframe.

- About 11 jobs run in parallel.

- These jobs manage to archive about 1.3 terabytes in about 20 hours.

- This process will soon be migrating to a 6400 silo.

While our archives are produced by a number of Database Administrators and System Administrators who perform different duties on each platform or mainframe, we are developing processes to migrate to Veritas. The primary business reason for the move is to centralize and better control the warehouse BAR.

Mark Bentlage
Senior Database Administrator
Development & Test Systems
Enterprise Data Warehouse

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