Talk:Online analytical processing/Archives/2013

Older comments
What's the story about the text "Around 50% of Olap licences are never deployed"? -- Waveguy
 * A nice indicator of how tricky Olap is to set-up and use. TwoOneTwo

Reply to Jay - technically yes, in usage sometimes no. You can regard it as an affectation and change it if you want. Thanks for pointing up Waveguy's question TwoOneTwo
 * I think usage also should be given a priority. Can you show me an article (outside Wikipedia of course) where OLAP is mentioned as Olap ? I couldn't find any instances from Google because Google searches are case-insensitive. Jay 08:18, 3 Oct 2003 (UTC)

The article states that Oracle Express was released in 1970. This cannot be true as Oracle did not exist as a product until 1979. Could this be a typo for 1990? Geoff97 22:32, 2 Jan 2004 (UTC)

Express has been around since about 1970 (with IRI) but was only taken over by Oracle in about 1995. -- TonyP 13:17, 23 Jan 2005 (UTC)
 * The history I've heard was that it was originally an MIT research project in the early 70s (PhD project maybe?) and was then productized. I believe IRI acquired it sometime in the mid to late 70s - previously I believe it was owned by a company called Multidimensional Systems or something? Cchiappa 22:02, 11 January 2006 (UTC)


 * Express was started at MIT, the beginning is fuzzy. Eventually, it was commercialized by a Boston company called MDS (Management Decision Sciences) which sold it in the 70's and 80's on a Prime minicomputer. Express was more popular in marketing department than finance departments, who were more partial competitors IFPS and Comshare. MDS was acquired by IRI in the late 80's. IRI sold it to Oracle in about 1996 when it was the #1 OLAP tool and proceded to run it into the ground. By the way, this whole article, and the one on Business Intelligence, is a mess. Neil Raden 05:18, 20 September 2007 (UTC)

I would dispute the definition of OLAP as given here. I would guess Codd's paper has to be the true definition since it coined the term. However, multi-dimension data and indexes were in use long before that paper. It really is a different way of indexing/accessing the data (as distinct from relational), and sometimes even different storage too (as in MOLAP). It is not specifically a snapshot of any relational database. Also, products such as Holos demonstrated that multi-cube and joining were just as applicable to OLAP as SQL data -- TonyP 14:45, 25 Jan 2005 (UTC)
 * Holos was, I guess, the first ROLAP product. It was fatally flawed in so many ways, though. Anyway, whatever definition you use, OLAP is characterized by interactive navigation through data, performing calculations on the fly without specifc instructions. In other words, not adding up things that are not additive, like %'s, or headcounts or balances over time. Their basic orientation is hierarchical attributes describing numerical values. Codd didn't come up with the rules, he was paid by Arbor Software to publish them and many feel he disgraced himself in the process. Neil Raden 05:18, 20 September 2007 (UTC)

Actually, just for the archive, Holos was never a ROLAP product in the true sense. It was primarily a MOLAP, and secondarily a HOLAP ('hybrid'). Unfortunately, it was widely misunderstood and all-too-often compared against pure query engines --TonyP (talk) 20:27, 8 April 2009 (UTC)

Market Shares - Copyrighted information
The market share data that comes from OLAP Report is copyrighted by "Optima Publishing Ltd". I'm not sure if the use here qualifies as Fair use.


 * The page says the following: "This is copyright material. You can make brief references to it freely, with attribution, but not reproduce large sections or the entire article without permission from the publisher. You are free to link to this page without permission." I assumed that sdfsdf copying the names of the companies and their market shares would fit under "brief reference" and "not reproduce large sections". Should we contact the publisher to clarify this ? Wikiolap 20:17, 12 January 2006 (UTC)


 * I missed that note on the page. I would think it qualifies as brief.

Merging FASMI
I think it is OK to leave FASMI at its own page and not to redirect to OLAP page.
 * Except, with all due respect to Nigel Pendse, no one in the industry Uses FASMINeil Raden 05:18, 20 September 2007 (UTC)

OLAP Operation Types
It would be nice to have a listing of the OLAP operation types (e.g. slice/dice, roll-up, drill-down, pivot, etc.) I don't have a good definition of them in my head so I came here to refresh. I think they're one of the key properties of OLAP, no?

--Michaell 04:19, 13 October 2006 (UTC)

"See "OLAP cube" Kgrr (talk) 11:12, 17 March 2008 (UTC)

Examples
It would be nice to have some examples, of measures and values, for instance. It is a little difficult for non-experts to understand without them.

--Gilliancs 19:32, 17 February 2007 (UTC)


 * I agree. I found this introductory article to be very useful.  Read under "OLAP cube examples".
 * Kgrr (talk) 19:53, 29 February 2008 (UTC)

Change Name
Recommend changing acronym name of article to full name: "Online Analytical Processing", and redirect "OLAP" to this page. This is parallel in construction with the "Online Transaction Processing" article. Comments? SqlPac 14:52, 17 May 2007 (UTC)
 * I prefer to keep things the way they are, where Online Analytical Processing redirects to OLAP. Wikiolap 15:01, 17 May 2007 (UTC)
 * That being the case, then perhaps Online transaction processing should redirect to OLTP. I'm not too concerned with which way it's done - I'm more concerned with giving these topics parallel construction.  I'll drop a line on the OLTP talk page and recommend the reverse of this change recommendation and see what people think then.  Thanks! SqlPac 16:25, 17 May 2007 (UTC)


 * I believe the current name does not comply with "Avoid the use of abbreviations, including acronyms, in page naming unless the term you are naming is almost exclusively known only by its abbreviation and is widely known and used in that form." Naming_conventions, and it is OLAP that should be moved. -- Beardo 01:46, 15 June 2007 (UTC)


 * That was my original thought as well, but with an objection on the table I don't have consensus. It looks like it was already done anyway, so until someone complains I guess it's good to go.  Thanks. SqlPac 04:54, 15 June 2007 (UTC)


 * It should stay this way. Consider if the acronym OLAP is used somewhere else, then it will become a disambiguation page.  It's very likely though that Online Analytical Processing will remain unambigous.Kgrr (talk) 20:02, 29 February 2008 (UTC)

WP:EL
Wikipedia is not a collection of links. The external references section becomes an open invitation to piles of SPAM if we are not careful. I will move good sources (Journal articles, books, and published magazine articles.) and delete the rest. I have copied the existing external references here in case there is a dispute:

(This link now (2008-03-06) resolves to http://www.oracle.com/technology/index.html, so don't expect to find Codd's paper there.) (found proper reference)
 * Providing OLAP to User-Analysts: An IT Mandate, the Paper by Codd first describing OLAP
 * OLAP Report: In depth overview of all commercial OLAP products (moved to references - FASMI)
 * OLAP Council White Paper (moved to references)
 * Introduction to OLAP (moved to references)
 * Dimensional Modeling and OLAP Tutorial
 * DataTime Universal: ROLAP product (moved to references)
 * Data Warehousing and OLAP: A Research-Oriented Bibliography (moved to bibliography)
 * Microsoft OLAP information
 * A chapter from Erik Thomsen's book OLAP Solutions: Building Multidimensional Information Systems, 2nd Edition
 * OLAP.com
 * Business Intelligence OLAP

I will help you get the article in shape. We just need to follow the guidelines. Kgrr (talk) 13:21, 17 March 2008 (UTC)

In essence, each action of slicing and dicing is equivalent to adding a "WHERE" clause in the SQL statement.
... is outright wrong for Dicing. Dicing refers to orienting the hypercube in another way, which basically means, that the usual matrix display changes axes, which often implies a different list of columns.

Slicing (setting filters), is equivalent to adding a "WHERE" clause. dicing is equivalent to changing the selected columns. — Preceding unsigned comment added by 80.130.29.135 (talk) 20:06, 7 June 2012 (UTC)

Merger proposal
I am proposing that OLAP cube be merged here. There is significant overlap to the extent that some of the other langauge Wikipedias are wikilinked to one or the other as if they were the same thing. I am probably doing this wrong, but I am more of a language/ general knowledge person, I don't really get too involved with the WP guts, so please, those who are better at this kind of thing, please weigh in. Best regards, Rui &#39;&#39;Gabriel&#39;&#39; Correia (talk) 18:59, 15 August 2012 (UTC)

Proposed Clarification
The terms measurement (often called a metric) and dimensions seem to be used differently here than they are in mathematics. The concepts of a measurement (metric) and a dimension are interchangeable in mathematics. Any measurement can be treated as a dimension of any hypercube within any vector space. This article makes a distinction between these two terms that should be clarified to help the formal mathematics students among the readership.

In mathematics there is often a distinction between independent and dependent variables, dimensions, or metrics. The terms range (set of possible values of the independent variable) and domain (set of possible values of the dependent variable) of a function or system is often used in mathematics as well. This may be what is intended for the dinstinction between measurement and dimension in this article. It seems that wherever this article says "measurement" the mathematics reader should think "dependent variable or dimension". And whereever the article says "dimension" the mathematics reader should think "independent variable or dimension"

Am I mistaken? Would it not be helpful for me to clarify the ambiguous business language of this article with more precise mathematics terms? — Preceding unsigned comment added by 70.66.13.60 (talk) 16:44, 1 September 2012 (UTC)


 * I'm not sure where the specific references are that you're mentioning. However, "measures" is usually a dimension consisting of multiple fields, each of which is a distinct measurement of something. Products vary in how similarly or differently they treat their dimensions. For instance a "time" dimension usually has properties that do not apply to others - such as an implicit ordering of the fields. The Holos product used to use the term "orthogonality" in its marketing since (at the time anyway), it treated its dimensions in an unusually consistent fashion. TonyP (talk) 14:06, 2 September 2012 (UTC)


 * Business jargon can be a bit quirky but, from a mathematical perspective, you can think of "measures" as the dependant variables. The relationship between variables and dimensions can be a little confusing: In a typical cube there is one dimension of "measures" in which each of the dependent variables is an "element" of the dimension, and each independent variable is a separate dimension in which the "elements" of the dimension are the discrete values which the variable can take. E.g., in a meteorological table, dependent variables like Rainfall or High_Temperature would each be "elements" of a single "measures" dimension, and independent variables like Date or City would each be a dimension unto itself, with discrete values like January or Paris as "elements". ~ Ningauble (talk) 18:47, 2 September 2012 (UTC)

Split out "Multidimensional databases"?
Multidimensional databases seem to be a larger subject than OLAP in database models. -- Beland (talk) 01:08, 5 March 2013 (UTC)
 * Er, we also have Dimensional database as a redirect to Data warehouse, and they sound very similar. -- Beland (talk) 01:25, 5 March 2013 (UTC)