Archives of the TeradataForum

Message Posted: Sat, 30 Aug 2003 @ 08:01:13 GMT

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

Have used MS Analysis Server to build very large cubes from Teradata tables at a couple of sites. I was pleased that with some effort was able to abstract dimensional model over a Teradata ER data warehouse using views to build cubes and avoid re-engineering the data architecture or excessively duplicating data in data marts or summary tables.

Analysis Server w/ Teradata PROS:

- Very good scalability via partitions (fact table view that access over 200 GB base data)

- Ability to combine multiple facts with virtual cubes (e.g. Sales, Shipments Inventory, Payments, Budgets, etc....)

- Very powerful customization and analytics via MDX calculated members

- Tunable aggregations (MOLAP)

- Accessible APIs for cube maintenance (DSO)

- Accessible APIs for custom reporting ADO-MD and MDX

- Large number of 3rd party tools

- Relatively cheap

- Reasonable drill-through capability (to base tables)

- Widespread use (equal 1st with Essbase in market according to OLAP Report)

Analysis Server w/ Teradata CONS:

- Windows only

- IIS only (does support XML for Analysis but this is a bit of work to get it going from Apache et. al.)

- Support for Teradata is via default ODBC or OLE-DB interface

- Requires some significant trial and error to get it to scale with Teradata

- Use with ROLAP option is problematic - no Teradata optimizer awareness

- Imature administration compared to Teradata or other relational products (e.g. tuning, productions support, workload management, etc...)

Hope this helps.

Cheers,

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