What is the largest Oracle OLAP cube that is out there in terms of total size, number of dimensions, and metrics? The company I recently joined uses Microsoft SSAS for building cubes pulling data from an Oracle DB. The size of the Microsoft SSAS cube is 18 TB (Tera Bites). I am encouraging these people to use Oracle OLAP, but there is push back saying Oracle OLAP won't scale to the extent Microsoft SSAS has.
I need some real data to defend my case. Has any one moved from Microsoft SSAS to Oracle OLAP?
I don't know if you can get this kind of information on a public forum, so it may be best to contact your Oracle sales rep. At a public level I know of this recent benchmark.
In addition to the link that David provided, some more points are:
(1). With cube compression, it could be that the same 18 TB cube in MSAS is smaller in OLAP. Its hard to prove it without doing a test load.
(2). Microsoft does have a very feature-rich front-end to develop/build dimensions and cubes for MSAS, compared to Oracle OLAP's AWM.
Most of the work is handled in relational sql-views before loading dimensions and cubes in Oracle-OLAP. That is why there is relatively less functionality in AWM.
(3). Another thing I heard about in MSAS is "Scoped Assignments". I am not sure what it is, so I can't say how it can be done in Oracle-OLAP. If your customer currently use this MSAS feature, then you can post a question on this forum providing some details.
(4). MSAS does provide good Excel reporting. If your client will continue using Excel, then you should look at Simba's Oracle OLAP MDX plugin for Excel. Search this forum and you will find information about it, or goto http://www.simba.com/MDX-Provider-for-Oracle-OLAP.htm
(5). Writeback is another feature that can be done easily in MSAS. With Oracle-OLAP, you will have to come up with a custom (although simple) solution, where the data will be written into a table first and then loaded into Oracle-OLAP cubes. Its a very very quick process.
(6). There are more choices for reporting tools compared to MSAS. Since Oracle-OLAP cubes are queried through SQL SELECT statements, as long as a reporting tool can generate SELECT statements for an Oracle database, it can also work with Oracle-OLAP.
(7). Finally, it is much easier to federate (or combine) relational and multi-dimensional data together between Oracle-OLAP and Relational, as both are in the same database. So you may not have to store all the data into OLAP cube, what is currently loaded into MSAS cube.
Overall, its a slightly different way of doing things between OLAP and MSAS. But I am almost certain that with the power and scalability of Oracle database, Oracle-OLAP will perform better than MSAS. You just have to know how to convert those MSAS cubes/dimensions into OLAP cubes/dimensions.
we have clients managing 18TB OLAP data very complex environment with several dimensions containing values in the millions. Write back isn't a problem - (we do it with APIs from different clients - currently preparing of v2 fully integrated with 11g metadata.)
Otherwise - the guidance from David and Nasar is on the money. You could easily come up with a solution carrying that amount of data using a few different strategies. THe OLAP integration with relational basically opens the floodgates, both from a volume as well as integration and automation perspective.