Archives of the TeradataForum

Message Posted: Mon, 15 Apr 2002 @ 13:15:16 GMT

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

Rohit,

It is my opinion that Teradata made an incredibly poor decision in selling Teracube to Microstrategy a few years back. Since then, Teracube has been promptly shelved by Microstrategy. The only alternative to shipping data from Teradata to an OLAP Server is to implement a ROLAP solution.

Currently, the only 2 OLAP Servers which can work with Teradata via ROLAP are Oracle Express and Analysis Services. However, Oracle Express is being integrated into Oracle 11i and, therefore, will no longer be able to communicate with Teradata. Microsoft Analysis Services ROLAP performance and SQL optimization is very, very poor.

We are currently evaluating MS Analysis Services. From the beginning, Microsoft consultants were only interested in selling us on MOLAP (discouraging ROLAP.) When we pushed them further about ROLAP, they suggested we only consider ROLAP using SQL Server (to make use of the additional features only utilized by Analysis Services when reading a SQL Server database.) Note, eventhough Analysis Services doesn't require an installation of SQL Server 2000, Microsoft recommends that SQL Server 2000 is installed to store the OLAP Meta Data Repository (which otherwise is stored in an Access database.)

In my opinion, Teradata has limited its Customer base to two "OLAP" solution:

1 - Microstrategy - ("OLAP-like", not really OLAP) unless you are willing to spend addition money on licenses for another front-end reporting tool, you are limited to the Microstrategy front-end. Microstrategy requires that all data from all sources must exist in Teradata (limits flexiblility of giving cubes to users due to Network issues, etc.)

2 - MS Analysis Services - very, very poor ROLAP performance and SQL optimization. From a strategic stand-point, Microsoft has been pretty-much giving away MS Analysis Services for these last 2 or 3 years. Once a substantial user base is built and legal issues are resolved, who is to say that they will not try to pull the same stunt that Oracle played on us (i.e., limit ROLAP to SQL Server.)

With the exception of Microstrategy, all other reporting tool vendors have done a piss poor job of incorporating Teradata specific features into their SQL generators. The concept of a Tier 1 reporting partner doesn't seem to extend much further than the ability to connect to Teradata.

Teradata needs to do something...fast. OLAP reporting is quickly becoming the standard for companies with mulitple data sources (all stored at different levels) and don't have the budget to buy 50+ nodes. We will all be left with the "egg on our faces" unless Teradata takes note of the current shift in market direction. I, personally, would not want to be the one to tell my Business Client that they cannot have OLAP functionality because our relational database vendor doesn't think that OLAP reporting is relevant.

In our case, Teradata feels that our need for an OLAP engine is the result our or data model. For a bench-marking fee, they will validate the worthiness of our model before seriously considering their OLAP partnership pit-falls. Of course, the only type of data model which will warrant the Teradata Seal of Approval is a completely 3rd normal form model. I agree, 3rd nornal form is the way to go; however, based upon our user reponse time requirements and our existing hardware, 3rd normal form modeling will not work; therefore, we have a 3rd normal form database which is the source of mulitple dependent datamarts with conformed dimensions for reporting. If we were to report against the 3rd normal form database, we would need to buy lots of new hardware (fancy that) to meet current user response time requirements.

My 2 cents...
On Teradata's behalf, they are currently tossing this issue around in Rancho Bernando. The more feed-back they hear from the Customer, the better.

Claybourne Barrineau

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