Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Oracle9i: what`s included?

Re: Oracle9i: what`s included?

From: Mark Townsend <markbtownsend_at_home.com>
Date: Sat, 30 Jun 2001 18:17:01 GMT
Message-ID: <B76365AD.4D5D%markbtownsend@home.com>

in article 3b3dc121$1_at_netnews.web.de, Christian Atzert at c.atzert_at_web.de wrote on 6/30/01 5:08 AM:

> Hi everyone!
>
>
> I`m writing my thesis on the evaluation of BI-solutions which is why I need
> detailed informations on the stuff!
> What are the so called "OLAP-services" of 9i? Is it Express integrated into
> the
> server or is it a standard ROLAP-engine?
>
>
> Thanks for replies or good links on the subject.
>
>
> Christian

Well - sort of both. Basically, a DBA defines the meta data for the cube(s) they require in the Oracle9i database data dictionary. The database will then look at the definition (cubes, dimensions, and measures etc), the underlying schema for the base data, and do one or both of two things

  1. Provide a ROLAP type representation of the cube using relational techniques such as materialized views, grouping sets, bit mapped indexes etc
  2. If some or any part of the cube cannot be fully represented relationally (i.e a large multi-dimensionla space), and requires instantiation (i.e cannot be quickly generated on the fly), Oracle9i will generate an OLAP representation of the cube (also stored in the database) - using the same algorithms and techniques used by Oracle Express, but now built and managed within the Oracle9i database - basically the Oracle Express developers have added most of the Express capability into the Oracle9i engine.

When an application or end user requests some sort of cube operation through the Java OLAP interface the optimizer will determine if the question can be answered from the relational data and use SQL to do so. If the relational model of the cube cannot provide the complete answer, then some or all of the interim result set will be passed to the OLAP server engine in the database, which will then perform the rest of the operation - either on the fly, or from the OLAP cube information it has already generated from step 2) above.

Changes to the data or schema etc is integrated through the stack - so loading new data causes all the 'instantiations' of the data in the MV's and cubes to also be re-generated

The upshot of this all this is that the initial question is acutally a moot point - it's not ROLAP or OLAP - it's the best combination of both.

There is a high level whitepaper on this available at http://otn.oracle.com/products/oracle9i/pdf/o9i_olap_bwp.pdf Documentation is also available at
http://otn.oracle.com/docs/products/oracle9i/doc_library/901_doc/olap.901/a8 8755/toc.htm
http://otn.oracle.com/docs/products/oracle9i/doc_library/901_doc/olap.901/a8 6720/toc.htm
http://otn.oracle.com/docs/products/oracle9i/doc_library/901_doc/olap.901/a8 8756/toc.htm Received on Sat Jun 30 2001 - 13:17:01 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US