Monash Information Services -- Oracle/OLAP

"Curt Monash's publications provide unmatched insight into
technology and marketplace trends. I have read them avidly for over a decade."

--Larry Ellison, Chairman
and CEO, Oracle

"Curt Monash possesses the rare ability to distill the essence of technological issues into understandable terms. He is particularly adept at melding a firm's product positioning, corporate strategy, and valuation parameters into a concise and coherent framework upon which one can make an informed investment decision. He is a trusted resource."

Matthew P. Kaufler, CFA
Portfolio Manager
Clover Capital Management, Inc.

Note To Folks Looking for Monash University

How Oracle Does OLAP

Oracle has changed signals so often on the-products-formerly-known-as-Express that a lot of people are pretty confused. But in Oracle10g, the situation is pretty simple, at least on the database side. And it’s closely analogous to the way Oracle handles text data.

OLAP cubes are stored in LOBs (Large OBject data), inside the general Oracle10g database. The contents of such LOBs can be indexed by a descendant of the old Express data management technology, much as a collection of documents can be placed into LOBs and then indexed via the technology formerly known as Oracle ConText.

Then you can get at these OLAP cubes via SQL. Specifically, there’s a relatively new – and rarely used – SQL Model clause that permits OLAP calculations to be embedded in a normal SQL statement. Conceptually, this works against embedded OLAP cubes and regular relational tables alike, just as text search works against both real documents and ordinary character fields. Performance-wise, however, any Oracle user who makes significant use of this capability will probably want to store OLAP data in LOBs, using Oracle’s multidimensional indices.

Of course, the OLAP/text analogy isn’t perfect. The primary way to get at OLAP data stored in Oracle10g is still the Express language; Oracle has nothing analogous in the text area. And since SQL statements are likely to perform calculations on OLAP data in ways that don’t happen on text data, there are issues with combining information from multiple cubes that don’t arise in the document case. But as this technology matures, it’s likely that the arguments for keeping OLAP data in a general-purpose database will be just as valid as those for integrating text.

For more information, please contact Curt Monash.

To reach Monash Information Services by phone, please call 978-266-1815.